Text mining technologies applied to free-text answers of students in e-assessment: an experimental study in Greek

Text mining technologies applied to free-text answers of students in e-assessment: an experimental study in Greek | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Text mining technologies applied to free-text answers of students in e-assessment: an experimental study in Greek Angelos Charitopoulos, Maria Rangoussi, Dimitris Metafas, Dimitrios Koulouriotis This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4387141/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 17 Jan, 2025 Read the published version in Discover Computing → Version 1 posted 10 You are reading this latest preprint version Abstract Educational text mining is a rapidly growing field, thanks to the adoption of current probabilistic and machine learning algorithms. The current study focuses on student e-assessment through open-ended questions that require answers in the form of free text (student essays). Their analysis and evaluation are resource-demanding tasks for the instructor, even when supported by modern e-learning platforms. Topic modelling through the Latent Dirichlet Allocation algorithm is employed in an experimental setup, aiming to (a) extract meaningful topics from the body of pooled student answers (interpretable in the educational context of the course), (b) align the extracted topics to the ‘native’ internal structure of the body of texts, and (c) produce recommendations for the teacher in the form of alternative (meaningful) restructurings of the e-assessment units and consequently of the course content units. Quantitative and qualitative evaluation of the extracted topic models yield positive results for the first two aims, while at the same time, and regarding the third aim, the extracted topic models directly recommend for the teacher possible restructurings of the course content. These recommendations are of practical use for the teacher, especially when he/she seeks to restructure a course, either by shrinking or by expansion (fewer or more internal units). In conclusion, topic modelling opens a spectrum of possibilities for the teacher interested to explore ways to improve the structure and organization of his/her course. educational text mining topic modelling Latent Dirichlet Allocation e-assessment student essays recommendation for teachers Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Figure 15 Figure 16 Figure 17 Figure 18 Figure 19 1 Introduction Text (Data) Mining, a branch of the wider Data Mining area, is a collective term for sophisticated computational methods that analyze unstructured data in the form of texts to extract abstract information such as meaning, sentiments or opinions. Text Mining (TM) is today considered as a mature and yet impressively growing field [ 1 ]. Building on more than half a century of research and development, with milestones such as the advent of World Wide Web, the search engines on it (Google), Machine Learning and Deep Learning algorithms [ 2 ], Natural Language Processing (NLP) and commercial chatbots, and the current Large Language Models (LLM) such as OpenAI’s ChatGPT 3 and 4, TM has permeated a wide spectrum of fields such as business, e.g. [ 3 ], healthcare, e.g. [ 4 ], communication, e.g. [ 5 ], entertainment, e.g. [ 6 ] and education, e.g. [ 7 ]. The affordances of TM for education have early been realized, investigated and exploited to improve teaching and learning [ 8 ], gradually shaping Educational Text Mining (ETM) as a distinct research area. The abundance of educational data, produced in digital form by Virtual Learning Environments (VLE), Open Educational Resources (OER), forums/chats/blogs and online collaborative environments, etc., along with the current advanced machine/deep learning algorithms for natural language processing, text classification and topic modeling, have allowed ETM to complement and enhance results of Educational Data Mining and Learning Analytics with results of text analysis and understanding. The interested reader is addressed to [ 9 ] and references therein for a comprehensive review on ETM. In the context of TM, topic modeling has emerged as an attempt to extract hidden (latent) themes or topics from single texts or bodies of texts and to group texts under these topics, in an unsupervised way. To this end, topic modelling has followed two major paths, the statistical approach, via Latent Semantic Indexing [ 10 ],[ 11 ] and subsequently Probabilistic Latent Semantic Indexing, [ 12 ] and the machine learning approach, via Latent Dirichlet Allocation [ 13 ]. The emergence of topic modeling has promoted education research to a higher level of sophistication, abstraction and scope, regarding the kind of questions addressed, e.g., [ 14 ], [ 15 ], [ 16 ], [ 17 ],[ 18 ]. Since its introduction in 2003, Latent Dirichlet Allocation (LDA) algorithm has rapidly found extensive use in various fields, e.g. [ 19 ]. In the scope of ETM, LDA has already been successfully employed for mining of concept maps from academic articles [ 20 ], feedback for teachers on evaluation of teaching by students [ 21 ], topic extraction from OER [ 22 ], structuring of teaching texts [ 23 ], text summarization [ 24 ], student satisfaction analysis [ 25 ], student attitudes towards online laboratories [ 26 ], feedback for teachers on student performance [ 27 ], and many others. A common thread that connects ETM studies is their aim to improve teaching and learning in practice: we have to fully understand teaching and learning in order to make it better. Concrete results on this aspect is the ultimate goal, implying that intelligence extracted from educational texts through text mining / topic modeling should be actionable intelligence. Along that line, the need for automated and streamlined analysis and understanding of the various types of texts produced in educational contexts is evident today, due to the accelerating speed and volume of text generation in digital form. This target is being continuously sought by NLP technology. From its early stages [ 28 ], [ 29 ] to the current state-of-the-art [ 30 ], [ 31 ], NLP has created the high-quality text preprocessing foundation upon which current text mining / topic modeling algorithms working at the processing level have proliferated. The path is not yet fully automated, however; certainly, a human-in-the-loop is still necessary, even critical, for getting results – [ 32 ], [ 21 ] among others. Among the various research directions of ETM, an important one is student assessment through tests or assignments with open-ended questions that require answers in the form of free text. Evaluation and grading of such assignments is one of the most demanding and resource-consuming teacher tasks. It is not surprising, therefore, that considerable research effort has been dedicated to automate (parts of) that process and to evaluate the accuracy and efficiency of such automation, given the significance of (i) fair grading, in the case of summative assessment, or (ii) constructive feedback, in the case of formative assessment, for student progress. The automated assessment of short-answer tests [ 33 ] and the extraction of feedback from reflective student responses [ 34 ] are examples of problems where conspicuous results have been obtained through NLP and ETM. For longer texts, considerable progress is made in text summarization [ 35 ], [ 24 ] and text comparison and alignment [ 36 ]. Closely related to assessment of student answers is the automated generation of test questions within given topics, as an aid for the teacher [ 37 ]. The current study focuses on student assessment on the basis of student answers to open-ended questions, in the form of longer texts (essays). In contrast to existing research works, however, the aim is not to automate evaluation and grading or question generation. Here, the interest is on aiding the teacher evaluate the internal structure of his/her teaching content on the subject, as reflected in the structure of the corresponding evaluation assignments and questions therein, and possibly re-structure material and assignments jointly, towards a new organization consisting of more distinct and more internally coherent units. This is not an issue of better course material sequencing; rather, it is an issue of recognizing themes and subunits in the course content to increase coherence. Topic modelling is therefore a suitable approach, given its capacity to extract hidden topics and cluster texts (student answers, in our case) under each topic. Another aim of this study is to comparatively evaluate student answers by contrasting them to model answers prepared by the teacher for class-level feedback purposes. Comparison is attempted at the domain of top words representing each topic identified within the body of student answer texts on the one hand and the body of teacher model texts on the other hand. Incongruent topics, as revealed by disjoint top word sets, would alert the teacher that assignment questions in their current form lack clarity and comprehensibility and should be rethought and rephrased or more radically changed. In accordance with the above research aims, the following research questions (RQs) are posed: 1. RQ1: Can topic modeling cluster a body of texts consisting of student answers into ‘meaningful’ clusters, i.e., into clusters aligned to the inherent internal structure and entities of the specific body of texts? 2. RQ2: Can the set of top words in each topic, extracted through topic modeling of student answers, successfully ‘represent’ the subset of texts clustered under that topic, i.e., allow the synthesis of a name or title for the texts clustered under that topic that is meaningful in the educational context of the specific body of texts? 3. RQ3: (If RQ2 is answered to the positive) Is it possible to evaluate the quality of student answers by contrasting topics and top words extracted from them against those extracted from teacher-provided model answers? 4. RQ4: In what practical ways, other than those implied in the previous RQs, can topic modelling be exploited by the teacher to improve his/her teaching? In order to answer these RQs, an experimental study is carried out as described in the following sections. Answers are meant primarily as feedback or recommendation for the teacher, to help him/her (i) realize and identify strong and weak ties and relations between units and sub-units of the course material, as reflected in the structure and contents of the corresponding assignment of each unit, and (ii) evaluate the clarity and comprehensibility of assignment questions and proceed to change them, where necessary. Furthermore, the teacher may restructure and reorganize material and assignments and re-evaluate the results (the new student answers) using the same topic modelling approach until he/she achieves a desirable level of performance. In that sense, the proposed approach is a semi-manual support tool that can be used iteratively for the improvement of teaching. 2 Materials and Methods 2.1 Methodology Methodologically the present study is a quasi-experiment rather than an experiment because the data set used for analysis was collected during an existing semester-long graduate course not designed specifically for this study: (i) the texts produced by students already enrolled in this course constituted the sample of the study (convenience sampling) while (ii) the course regulation requires that all enrolled students be offered / taught the course in the prescribed way (no control group). Data collection, preprocessing and analysis were performed in batch mode, off-line, after the completion of the semester, as detailed in the following paragraphs. 2.2. Data Set Preparation The graduate course ‘E-learning Systems and Distance Learning Technologies’ of the Master Degree Program ‘ICT for Education’ was used as the source of texts for the present study. The 2022-23 spring semester class consisted of 19 students whose progress was evaluated through 6 assignments spread across the semester, one assignment every other week. Final course grade was the average of the (personal) best 5 out of 6 individual assignment grades. Each assignment consisted of a different number of questions depending on the subject, as detailed in Table 1 ; therefore, every student should have answered a total of 20 questions by the end of the course. E-assessment was carried out in the moodle e-learning platform, running on the departmental moodle server. There was no time limitation other than the due time for turning the answers in, which was uniformly set to 2 weeks from assignment time. Table 1 Number of questions per assignment Assignment 1 Assignment 2 Assignment 3 Assignment 4 Assignment 5 Assignment 6 Q.1.1 Q.2.1 Q.3.1 Q.4.1 Q.5.1 Q.6.1 Q.1.2 Q.2.2 Q.3.2 Q.4.2 Q.5.2 Q.6.2 Q.2.3 Q.3.3 Q.4.3 Q.5.3 Q.6.3 Q.5.4 Q.6.4 Q.5.5 For each subject treated in class, students were subsequently directed to study certain learning content made available on moodle, and then answer the corresponding assignment questions in the form of free-text answers, also in moodle. Grading and personalized feedback per question were also provided in moodle by the class instructors (2nd and 3rd authors of the current study). Furthermore, the class instructors composed and uploaded in moodle one set of ‘model’ answers to all questions of the current assignment, in order to provide feedback at a non-personalized, class level. This course organization has produced (19 students x 20 answers/student) + (1 instructor x 20 answers/instructor) = 400 texts, in total. The texts were extracted from moodle as 400 distinct _.txt files and renamed according to a convenient nomenclature as xyz.txt, where x = 1, ..., 6 denotes the assignment number, y = 2 or 3 or 4 or 5 denotes the number of questions within the specific (x-th) assignment, and z = 1, ..., 19 denotes the student ID. For example, 4.3.18.txt is the answer of student with ID 18 to the 3rd question within the 4th assignment. As the working language in this Master Degree Program is Greek, questions and answers were composed in Greek – a fact that proved to be a challenge in the subsequent steps of data preprocessing and data analysis. 2.3 Data Preprocessing RapidMiner is the environment used to implement the successive steps of data preprocessing and data analysis ( https://altair.com/altair-rapidminer ). Out of the variety of alternative environments, RapidMiner is selected thanks to its well-structured and self-evident interface, the wide spectrum of implemented algorithms, as well as the availability of sufficient documentation and help, as a result of its widespread use (REF). Certain limitations of RapidMiner regarding visualization of the analysis results, however, have led to the search for a complementary tool specialized in visualization, such as Alteryx ( https://www.alteryx.com/ ) or Orange ( https://orangedatamining.com/docs/ ). The later was selected because it can handle Greek language texts/words. Regarding data preprocessing, the typical sequence of steps is implemented in RapidMiner: data organization in 400 separate files, one file per answer text, named as described in the previous section and pre-stored in a local directory; data import to the RapidMiner environment in the form of a collection of text files; tokenization (breaking of each text into words); transformation into uniform-case fonts (here, uppercase); removal of stop-words (considered to carry no useful information as to the topic of the text); word stemming (removal of suffixes to reduce all inflexed forms of a given word into a single form); word filtering according to length in characters. Figure 1 and Fig. 2 illustrate the relevant sequence of steps as a RapidMiner two-level process. It should be noted here that, unfortunately, Greek is not among the languages currently supported by RapidMiner – or by any other equivalent environment, come to that. Special care has therefore been given to the language-dependent steps of stop-word removal: A custom-made dictionary of Greek stop-words was manually compiled in the form of a . txt file and used as a filter in RapidMiner to remove stop-words. stemming: During the last three decades intensive research in the field of Natural Language Processing for the Greek language has produced several stemmers, the major ones being TZK [ 38 ], AMP [ 39 ] and the Ntais Stemmer in its original and enhanced forms [ 40 ], [ 41 ], [ 42 ], based on the Porter stemmer algorithm [ 43 ]. The Greek stemmer more recently developed by [ 44 ] in the context of the Metaphor Detection project of NCSR ‘Demokritos’ ( http://metaphor.iit.demokritos.gr/ ) is employed here [ 45 ]. A Python implementation is available online in GitHub ( https://github.com/kpech21/Greek-Stemmer ), under GNU General Public License. This code was inserted in RapidMiner and executed on an installation of Python 3.10, using the ‘Execute Python’ operator, to carry out stemming per text file. An excerpt of the python script executed in RapidMiner is shown in Fig. 3 . 2.4 Data Analysis Topic modeling through the LDA algorithm [ 13 ] is employed in order to answer the two research questions defined at the outset of this study. The essential element in LDA is the introduction of an intermediate layer of ‘hidden’ (latent) variables (the topics) between the apparent entities at the higher level (texts) and those at the lower level (words). LDA is implemented in RapidMiner (Fig. 4 ). A major issue in topic modeling, common in all unsupervised methods, is the decision on the optimal number of topics, k , which is required as an input parameter by all relevant algorithms. The same holds true for clustering algorithms that operate in the domain of numerical or categorical data instead of text data. For the latter case, k is selected through various alternative methods, such as the Silhouette index [ 46 ], the Davies-Bouldin index [ 47 ] or the Calinski-Harabasz index [ 48 ]. For the case of text data, the optimal k is usually sought by minimizing the perplexity or maximizing the loglikelihood of the clustered data set, across a range of values of k . Decisions that favor empirically set values of k , lower than the optimal value obtained through optimization, when the latter is impractically or unreasonably high given the application, are also described, e.g., [ 49 ]. In the present case, however, k is considered as a known parameter: depending on whether the texts in the current data set are clustered (i) as to the assignment they answer to ( k = 6 ) or, (ii) as to the question within the assignment they answer to ( k = 20 ), or (iii) as to the ID of the student who authored them ( k = 19 ). The optimization step is therefore not necessary, in either of these cases. Similar, application-driven rather than optimization-driven selections of k , are described, e.g., by [ 23 ] or by [ 27 ]. It does remain meaningful, however, to perform a search in the vicinity of those initial values of k , for values that represent a ‘better’ or more ‘meaningful’ clustering of the texts, and then try to interpret this ‘better’ clustering and discuss its implications for the teacher. Consequently, experimentation is organized into (i) a preliminary experiment, (ii) two major clustering experiments where k is known (given), and (iii) a final, exploratory experiment where a range of k values are investigated. In the following paragraphs, experiments are described and results are reported jointly for each part. A common thread across all experiments is the problem of content-based clustering of a body of texts, where ‘content’ is represented by the topics extracted by LDA and the top words identified in each topic. The capacity of LDA to discern the internal structure inherent in a body of texts and to align the extracted topics to the native entities of this structure is investigated experimentally on the current data set. The (known) internal structure of the data set into assignments and further into questions facilitates the evaluation of the results and consequently the validation of LDA-derived topic models. 3 Results 3.1 Preliminary experiment – total body of texts The preliminary experiment aims for a first, rough estimation of the capacity of LDA-based topic modelling to accurately represent the overall course contents. To this end, all 380 student texts are taken in a single cluster ( k = 1 ) and fed to LDA in order to extract the top 10 words to characterize it. For comparison purposes, the 20 teacher texts, again as a single cluster ( k = 1 ) are also fed to LDA for the extraction of the corresponding top 10 words. Results are given in Table 2 and visualized in Fig. 5 , where words are in Greek, stemmed and accompanied by an English translation in brackets, for readers’ convenience. A qualitative evaluation of these results is performed by the 2 class instructors who, in the role of field experts, independently reviewed these two lists of top 10 words and scored their relevance to the theme of the course, on a 5-level Likert scale of relevance: {0 = irrelevant; 1 = loosely relevant; 2 = moderately relevant; 3 = closely relevant; 4 = fully relevant or identical}. Final relevance scores, also shown in Table 2 (bottom line), are calculated as the average of the two independent scores, where they do not differ by more than 2 levels in the scale; cases where independent scores differ by 3 levels or more are discussed and unanimously agreed scores are given. Table 2 Top 10 words in the total body of student texts (380) and teacher texts (20), and relevance scores a In Student Texts (380) weights normalized weights In Teacher Texts (20) weights normalized weights µαθησιακ 567 1.00 εκπαιδευοµεν 22 1.00 µαθητ 475 0.84 µαθησιακ 19 0.86 εκπαιδευτ 466 0.82 συγκεκριµεν 19 0.86 συστηµ 348 0.61 µαθης 18 0.82 µαθης 339 0.60 γνως 17 0.77 γνως 333 0.59 αποτελεσµ 17 0.77 στοχ 312 0.55 επιπεδ 15 0.68 εκπαιδευοµεν 299 0.53 χρηστ 13 0.59 τροπ 279 0.49 συστηµ 12 0.55 αποτελεσµ 263 0.46 εκπαιδευτ 12 0.55 RELEVANCE 4 RELEVANCE 4 a shaded entries: common top words between student answers and teacher model answers. As it can be observed in Table 2 and Fig. 5 , 7 out of the 10 top words (70%) are common in student answers and teacher model answers although not in the same ranking. Figure 5 reveals the strong alignment of terms employed by the students and the teachers; the fact that they do not fully coincide is welcome as it means that students do not replicate expressions and vocabulary found in the study material provided by the teacher; rather, they compose their own texts. Relevance scores indicate ‘full relevance’ (4 / 4) both for the student top words and for the teacher top words. In fact, either of the two lists could help a researcher compose a title for this course ‘in blind’, having to do with “ the effectiveness of (e-)learning-based educational systems regarding cognitive / learning gains of the learners ” – which is a satisfactory summary of the actual course content. 3.2 Clustering of Texts Through LDA – Assignments Level This first clustering experiment views the data set as a collection of 6 distinct subsets of texts, each subset consisting of all answers to the questions in a single assignment. Clustering of the 380 student texts under k = 6 LDA-derived topics is intended to investigate the potential of LDA to align the extracted topics to the inherent internal structure of the total body of texts, i.e., the assignments it consists of. The main processing step (Fig. 4 ) is therefore executed once, with k = 6 , LDA parameters a and b empirically set to a = 0.1 and b = 0.01 and 2,500 iterations. Ideally, each LDA-derived topic should uniquely identify with one of the 6 assignments, i.e., LDA should be able to (i) cluster under each of the identified k topics those and only those student texts that answer questions in a single assignment; (ii) extract ‘meaningful’ top words in each topic that accurately represent the theme assessed by the corresponding assignment. Accordingly, LDA performance in these two tasks is evaluated regarding clustering: quantitatively, through accuracy (%) of the classification task, calculated as the percentage of the main diagonal values over the total values in the relevant 6 x 6 confusion matrix of topics v/s assignments; regarding top words: quantitatively, by the percentage of common top words between (a) the top 10 words in each LDA-extracted topic and (b) the top 10 words in the corresponding assignment, as these are extracted by running LDA independently 6 times, each time on the subset of student texts known to belong to the i -th assignment ( i = 1, …, 6 ); regarding top words: qualitatively, by the 2 class instructors who, in the role of field experts, reviewed the top 10 words extracted by LDA in each topic and scored their relevance to the theme of the corresponding assignment, on the same 5-level Likert scale adopted for the preliminary experiment and following the same procedure (averaging). The same procedure of relevance scoring they performed on the top 10 words derived by LDA independently in each assignment, the latter considered as a ‘benchmark’ for the former. For comparison purposes, the 20 teacher model answers are also pooled in a single body of texts and fed to LDA for topic modeling into k = 6 topics and the extraction of top 10 words in each topic. The experimental setup of this first clustering experiment is clarified in Fig. 6, where continuous lines are used to frame known/given units (6 assignments), while dashed lines are used to frame LDA-extracted units (6 topics). Quantitative evaluation results are given in Table 3 .a [3.b] for student [teacher] texts, and illustrated in the corresponding Fig. 7 .a [7.b]. Table 3 a Student texts: confusion matrix of 6 assignments versus k = 6 LDA-derived topics 1 (topic_5) 2 (topic_4) 3 (topic_1) 4 (topic_3) 5 (topic_2) 6 (topic_0) Assignm.1 38 / 38 100.00% Assignm.2 57 / 57 100.00% Assignm.3 1 / 54 53 / 54 1.85% 98.15% Assignm.4 56 / 57 18 / 57 63.16% 31.58% Assignm.5 1 / 95 38 / 95 56 / 95 1.05% 40.00% 58.95% Assignm.6 18 / 76 58 / 76 23.18% 76.30% Accuracy 66.78% Table 3 b Teacher texts: confusion matrix of 6 assignments versus k = 6 LDA-derived topics 1 (topic_2) 2 (topic_5) 3 (topic_4) 4 (topic_1) 5 (topic_0) 6 (topic_3) Assignm.1 2 / 2 100.00% Assignm.2 1 / 3 2 / 3 33.33% 66.67% Assignm.3 3 / 3 100.00% Assignm.4 2 / 3 1 / 3 66.67% 33.33% Assignm.5 2 / 5 3 / 5 40.00% 60.00% Assignm.6 1 / 4 3 / 4 25.00% 75.00% Accuracy 78.06% Results indicate that content-based clustering of texts according to LDA-derived topics is both feasible and meaningful in the educational context of the current data set. Indeed, and despite the non-ideal accuracy calculated for clustering student texts in Table 3 .a (66.78%) or teacher texts in Table 3 .b (78.06%), these percentages are significant and not compatible with ‘randomness’ in the results. On the other hand, the off-diagonal entries in either of the two confusion matrices indicate to the teacher possible rearrangements across assignments, for a more coherent overall structure, as discussed in the Discussion section below. Regarding the top 10 words extracted from each topic, these are shown in Table 4 , in Greek, stemmed, and sorted in descending order of weights. Top words extracted from student texts are shown in the left-hand side of Table 4 while those extracted from teacher texts are shown in the right-hand side of Table 4 , for comparison. The columns labeled as ‘Topic_i’ ( i = 1 .. 6 ) present the top words extracted by LDA from each LDA-extracted topic (dash-lined frames in Fig. 6), while the columns labeled as ‘Assignment_i’ ( i = 1 .. 6 ) present the top words extracted by LDA from each given subset of texts belonging to this assignment (continuous-lined frames in Fig. 6). The latter, therefore, act as ‘benchmarks’ for the former. The Topic_2 zone of Table 4 is visualized in Fig. 8 , to compare LDA-derived results between students and teacher (‘Topics’ columns only). Each of the 3rd and 6th columns give the number of common top words over 10, between the two preceding columns. These number are averaged across the 6 groups in the bottom line of Table 4 . For student texts, common top words between given groups of texts (assignments) and LDA-derived groups of texts (topics) range from 2 / 10 (Topic_1 versus Assignment_1) to 10 / 10 (Topic_3 versus Assignment_3) with an average of 5.83 / 10 or 58.3%. For teacher texts, the range is 6 / 10 (Topic_1 versus Assignment_1) to 9 / 10 (Topic_3 versus Assignment_3) with an average of 7.16 / 10 or 71.6%. The fact that both these similarity scores are above 50% is satisfactory. At the same time, this quantitative evaluation is in good agreement to the accuracy results obtained in Table 3 .a (66.78%) for student texts and in Table 3 .b (78.06%) for teacher texts, where again teacher text scores are higher than student text scores. A result that is probably more informative as to the quality of the obtained topic models is the percentage of common words • between student texts (‘Topics’ column) and teacher texts (‘Topics’ column): this ranges from 0 / 10 to 6 / 10 with an average of 3.17 / 10 or 31.70%. • between student texts (‘Assignments’ column) and teacher texts (‘Assignments’ column): this ranges from 2 / 10 to 6 / 10 with an average of 4.33 / 10 or 43.33%. These results are satisfactory, although somewhat lower when the units are extracted by LDA (‘Topics’ columns) than when the units are given (‘Assignments’ columns). Table 4 Student texts (left) and teacher texts (right): top 10 words of the LDA-extracted k = 6 topics and of the 6 assignments a Student Texts Common words Teacher Texts Common words Topic_1 Assignment_1 2 Topic_1 Assignment_1 6 µαθησιακ εκπαιδευτ εκπαιδευοµεν µαθηµ θεωρ ταξινοµ µαθηµ εκπαιδευοµεν τροπ µαθητ βελτιως επιπεδ µαθης γνωστ προγραµµ γνως µαθητ στοχ µαθης βελτιως ερευν γνως επιπεδ αρχ ατοµ µαθησιακ εκπαιδευτ ταξινοµ αποτελεσµ διαδικας γνως σηµαντικ πληροφορ αξιολογης αρχ διαστας µοντελ δηµιουργ συγκεκριµεν προγραµµ Topic_2 Assignment_2 8 Topic_2 Assignment_2 8 µαθησιακ µαθησιακ γνως γνως εκπαιδευτ αποτελεσµ αποτελεσµ αποτελεσµ αποτελεσµ γνως διαδικας µαθησιακ γνως εκπαιδευοµεν θεωρ µαθης ταξινοµ µαθης σκεψ περιγραφ αξιολογης εκπαιδευτ πληροφορ θεωρ µαθης ταξινοµ ταξινοµ επιπεδ δηµιουργ αξιολογης µαθης διαδικας στοχ προγραµµ επιπεδ ταξινοµ εκπαιδευοµεν διαδικας πρακτ δεξιοτητ Topic_3 Assignment_3 10 Topic_3 Assignment_3 9 ιστοσελιδ ιστοσελιδ αισθητ καλ χρωµ χρωµ κανον περιεχοµεν πληροφορ πληροφορ ιστοτοπ αισθητ κανον σελιδ περιεχοµεν κανον σελιδ κανον καλ ιστοτοπ καλ καλ πολλ πολλ κειµεν χρηστ κακ κακ χρηστ περιεχοµεν χρωµατ χρωµατ σηµαντικ σηµαντικ σελιδ σελιδ περιεχοµεν κειµεν µεγαλ πληροφορ Topic_4 Assignment_4 6 Topic_4 Assignment_4 7 µαθησιακ µαθητ προσαρµοστ τεχνολογ συστηµ συστηµ εκπαιδευοµεν προσαρµοστ εκπαιδευτ εκπαιδευτ τεχνολογ εκπαιδευοµεν προσαρµοστ προσαρµοστ µοντελ ασκης στοχ µαθησιακ συστηµ µαθητ µαθητ στοχ ασκης συγκεκριµεν εκπαιδευοµεν γνως χρηστ µοντελ χρηστ απαντης συγκεκριµεν προβληµ διαφορετ µαθης υποστηριξ λυς µαθηµ προβληµ πλοηγης ευφυ Topic_5 Assignment_5 2 Topic_5 Assignment_5 6 µαθητ µαθησιακ µαθησιακ µαθησιακ προβληµ εκπαιδευτ µαθητυπ συγκεκριµεν απαντης µαθητ ερευν εκπαιδευοµεν συστηµ συστηµ ατοµ µαθητυπ τεχνολογ τροπ εκπαιδευς χρηστ λαθ εκπαιδευοµεν αποτελεσµ προσαρµοστ επιλυς συγκεκριµεν συγκεκριµεν τροπ λυς µαθης τροπ ατοµ µοντελ στοχ εκπαιδευτ αποτελεσµ προσφερ διαφορετ εννοι µοντελ Topic_6 Assignment_6 7 Topic_6 Assignment_6 7 συναισθηµατ συναισθηµατ συναισθηµατ συναισθηµατ συναισθηµ συναισθηµ χρηστ χρηστ υπολογιστ συστηµ τεχνολογ µαθης συστηµ υπολογιστ συστηµ τεχνολογ καταστας καταστας καταστας συστηµ εκφρας µαθητ µαθης καταστας αναγνωρις µαθησιακ συνθες συναισθηµ χρηστ µαθης φων συνθες ανθρωπιν εκφρας µηνυµ ληψ µοντελ µοντελ ευνοικ αποφας Average common words 5.83 / 10 Average common words 7.16 / 10 a shaded entries: common top words between student answers and peer teacher answers. Table 5 summarizes all relevance scores for top word lists from student texts and from teacher texts, as given by the 2 class instructors in the adopted 5-level scale. The columns labeled as ‘Topics’ and ‘Assignments’ in Table 5 are aligned with the corresponding columns in Table 4 . These qualitative evaluation results indicate that LDA-derived topic models as represented by the respective top 10 words are in good alignment to the educational context of the specific data set. Indeed, relevance scores in Table 5 are in the three higher levels of the 5-level scale (0 to 4), with an average relevance of 3.33 / 4.00 or ‘closely relevant’, for words in LDA-derived topics from student texts (2nd column) and in LDA-derived topics from teacher texts (4th column) alike. Moreover, average relevance scores when units (assignments) are given are also close: 3.42 / 4.00 for student texts – 3rd column against 3.58 / 4.00 for teacher texts – 5th column. The facts that (i) individual entries as well as averages in adjacent columns under ‘Student Texts’ are close, (ii) the same holds true for adjacent columns under ‘Teacher Texts’, while (iii) neither of the lists has any ‘irrelevant’ or ‘loosely relevant’ cases, are welcome since any significant difference in relevance scores, or numerous ‘irrelevant’ or ‘loosely relevant’ cases, would question the validity of the models. Furthermore, the fact that average relevance scores when the units are LDA-extracted (topics, 2nd and 4th columns) are only a little lower than when units are given (assignments, 3rd and 5th columns) is a direct result of the good quality of LDA clustering, as summarized in Tables 3 .a and 3.b. It can therefore be claimed that quantitative and qualitative results are in good agreement. Table 5 Relevance scores for top 10 words in student texts and in teacher texts, across the 6 LDA-derived topics and the 6 given assignments Student Texts Teacher Texts nr. Topics Assignments Topics Assignments 1 3.00 3.50 2.50 3.50 2 3.50 4.00 3.00 4.00 3 4.00 3.50 4.00 3.50 4 3.50 3.00 3.50 3.50 5 2.50 3.00 3.50 3.50 6 3.50 3.50 3.50 3.50 Average 3.33 3.42 3.33 3.58 3.3 Clustering of Texts Through LDA – Questions Level In the 2nd clustering experiment, the data set is considered as a collection of 20 distinct subsets of texts, each subset consisting of all student answers to a single question (note that questions across all 6 assignments add up to 20). Clustering of the 380 student texts under k = 20 LDA-derived topics is intended to investigate the potential of LDA to align the extracted topics to the inherent internal structure of the total body of texts, i.e., the questions it consists of. This is a finer level of analysis compared to the assignments level investigated in the 1st experiment. The main processing step (Fig. 4 ) is therefore executed once, with k = 20 , LDA parameters a and b empirically set to a = 0.1 and b = 0.01 and 2,500 iterations. For comparison, the 20 teacher model answers are also pooled in a single body of texts and fed to LDA for topic modeling into k = 20 topics. The setup of this 2nd experiment is clarified in Fig. 9, where continuous lines are used to frame known/given units (questions) while dashed lines are used to frame units extracted by LDA (topics). Performance of LDA models in this second experiment is evaluated quantitatively through confusion matrices and accuracy (%) scores and qualitatively, through top word comparison and relevance scores, as in the first experiment. Quantitative results are shown in Table 6 .a [Table 6 .b] and visualized in Fig. 10 .a [Figure 10 .b] for the student [teacher] texts. Accuracy calculated from the confusion matrix in Table 6 .a [Table 6 .b] is 66.93% [33.50%] for the student [teacher] texts clustered into 20 LDA-derived topics. Accuracy of student texts (66.93%) is at the same level with the accuracy obtained in the 1st experiment, for clustering of the same texts into k = 6 topics intended to align LDA-derived topics to assignments. The fact that performance is sustained when the number of topics k (and, therefore, the complexity of the task) is dramatically increased from 6 to 20, is a strong indication that the adopted approach and the LDA-derived topic models are both feasible and meaningful in the educational context of the current data set. On the other hand, accuracy of teacher texts drops to 33.50% for k = 20 topics, as compared to the 78.06% obtained for k = 6 topics in the first experiment. Closer inspection of Fig. 10 .b reveals that this is due to the fact that LDA optimally clusters teacher texts under 7 (practically, under 6) topics, despite the ‘availability’ of more topics when k is set to 20. It is interesting that results are practically unaltered (7 topics, accuracy 40.0%) when LDA is re-run with a lower parameter a = 0.01 , favoring an increased number of topics. This result is a strong indication for the teacher on the question of whether to further break down the course content to more than 6 units (assignments) or not, as discussed in the Discussion section below. Top 10 word lists for the 20 topics are not listed in full for this 2nd experiment, as was the case for the 1st experiment (Table 4 ), because of their volume. Instead, Table 7 shows the corresponding information of Table 4 for a single characteristic example among the 20 LDA-derived topics, namely, Topic_14, manually matched to question Q1.1 for the student texts and Topic_4, manually matched to Q1.1 for the teacher texts. The 3rd and 6th columns give the number of common top words over 10, between the two preceding columns, which are 7 / 10 for the student list and 8 / 10 for the teacher list. All common top word counts are averaged across the 20 lists and the results are shown in the bottom line of Table 7 . For student texts, common top words between given groups of texts (questions) and LDA-derived groups of texts (topics) range from 0 / 10 (Topic_18 versus Q3.3) to 10 / 10 (Topic_19 versus Q2.2) with an average of 5.60 / 10 or 56.0%. For teacher texts, the range is 2 / 10 (Topic_7 versus Q5.1) to 8 / 10 (Topic_4 versus Q1.1) with an average of 6 / 10 or 60.0%. The fact that both these similarity scores are above 50% is satisfactory. At the same time, this quantitative evaluation is in good agreement to the accuracy results obtained in Table 3 .a (66.78%) for student texts and in Table 3 .b (78.06%) for teacher texts, where again teacher text scores are higher than student text scores. A result that is probably more informative as to the quality of the obtained topic models is the percentage of common words (i) between student texts (‘Topics’ column) and teacher texts (‘Topics’ column): this ranges from 0 / 10 to 5 / 10 with an average of 2.86 / 10 or 28.6%. (ii) between student texts (‘Questions’ column) and teacher texts (‘Questions’ column): this ranges from 1 / 10 to 6 / 10 with an average of 3.60 / 10 or 36.0%. These results are lower than the corresponding ones in the first experiment (31.70% and 43.33%, respectively) – a direct impact of the increased complexity of this task ( k = 20 instead of k = 6 ) on the results. As in the 1st experiment, results are somewhat higher when the units are given (‘Questions’ columns) than when the units are extracted by LDA (‘Topics’ columns), which is an expected outcome. Table 7 Student texts (left) and teacher texts (right): a single example of a top 10 words list out of the 20 similar lists, comparing top 10 words of LDA-extracted k = 20 topics against top 10 words of the question manually matched to each topic a Student Texts Common words Teacher Texts Common Words Topic_14 Question_1.1 7 Topic_4 Question_1.1 8 ταξινοµ εκπαιδευτ εκπαιδευοµεν µαθηµ εκπαιδευτ ταξινοµ γνως εκπαιδευοµεν αξιολογης στοχ µαθηµ προγραµµ γνωστ µαθησιακ προγραµµ επιπεδ διαδικας µαθητ επιπεδ γνως γνως αξιολογης βελτιως βελτιως επιπεδ συγκεκριµεν αρχ αρχ στοχ µαθηµ ταξινοµ συγκεκριµεν συγκεκριµεν γνωστ οργανως ταξινοµ προγραµµ προγραµµ µαθησιακ στοχ … … … … … … … … … … … … Average common words 5.60 / 10 Average common words 6.00 / 10 a shaded entries: common top words between student answers and peer teacher answers. Qualitative results are based on the relevance scores given by the 2 class instructors who assessed the relevance of each of the 20 top word lists obtained from LDA-derived topics to the theme of the question manually matched to the specific topic. Results are gathered in Table 8 for student texts (left side) and teacher texts (right side). Table 8 Relevance scores for top 10 words in student texts and in teacher texts, across the 20 LDA-derived topics and the 20 given questions (ordered by Question numbers) Student Texts Teacher Texts nr. Topics Questions Topics Questions 1 (Q1.1) 3.5 3.5 4.0 3.5 2 (Q1.2) 1.5 2.5 3.0 3 (Q2.1) 3.5 4.0 3.5 4 (Q2.2) 3.0 2.5 2.5 3.0 5 (Q2.3) 3.0 3.0 3.0 6 (Q3.1) 3.5 3.5 4.0 3.5 7 (Q3.2) 2.5 2.5 2.5 8 (Q3.3) 2.5 3.0 3.0 9 (Q4.1) 3.0 4.0 3.0 2.5 10 (Q4.2) 4.0 2.5 3.5 11 (Q4.3) 3.0 3.5 4.0 12 (Q5.1) 3.5 4.0 2.0 3.5 13 (Q5.2) 3.5 3.5 3.5 14 (Q5.3) 3.0 2.5 3.0 15 (Q5.4) 3.0 2.5 2.0 3.0 16 (Q5.5) 3.5 2.5 3.0 17 (Q6.1) 4.0 4.0 3.5 18 (Q6.2) 3.0 4.0 2.5 2.5 19 (Q6.3) 3.0 3.5 3.0 20 (Q6.4) 3.5 2.0 2.5 Average 3.15 3.15 2.86 3.13 Relevance scores in Table 8 are at the upper 3 levels of the (0–4) scale, indicating the high quality of the 20 LDA-extracted topics and their good correspondence to the 20 questions in this course. In the student texts, average relevance scores range from 2.0 to 4.0 with a mean value of 3.15 / 4.00 or ‘closely relevant’, obtained for LDA-extracted units (‘Topics’ column) and for given units (‘Questions’ column) alike. None of the cases is marked as ‘irrelevant’ or ‘loosely relevant’ – a welcome result, since lower relevance scores with numerous ‘irrelevant’ or ‘loosely relevant’ cases, would question the validity of the LDA-derived models. It is interesting and worth reporting that a comparison of the top 10 student words in Table 8 to the corresponding top 10 teacher words has not been possible. This is a direct consequence of the relevant comment made in the quantitative results of the k = 20 experiment (previous paragraph). There, it was detected that when LDA runs on the body of 20 teacher texts with k = 20 , it identifies and ‘populates’ with top words only 7 topics, while the rest up to 20 are not exploited and remain empty of top words (4th column). For those 7 topics extracted from teacher texts and populated with top 10 words (4th column), relevance scores are close to the corresponding ones when the units are given (5th column), while their average across all 7 lists is 2.86 (‘closely relevant’) that is a little lower than the average across all 20 (given) units in the 5th column (3.13, ‘closely relevant’). 3.4 Clustering of Texts Through LDA – Search and optimization in the vicinity of k = 6 topics This last, exploratory experiment investigates the robustness of the structure of the course for certain alternative numbers of internal units ( k values). In the 2nd clustering experiment, where k was set to 20, the 20 teacher texts (one per question) remained clustered under 6 or 7 topics, when more clusters were available. This inspired us to perform a search in the vicinity of k = 6 for values of k that might result in more coherent structures. Practically, the 1st clustering experiment with k = 6 is repeated here in the vicinity of 6, namely, for each k in the set {4, 5, 6, 7, 8}. Calculated perplexity values, shown in Fig. 11 as a function of k , verify that k = 5, 6 , 7 represent lower perplexities than k = 4 or k = 8 , possibly leading to more coherent structures. Although the current structure of k = 6 does not represent the local minimum in this range, the previous ( k = 5 ) and the next ( k = 7 ) values are fairly close and practically equivalent. This result suggests that it is worth assessing the plausibility of either of these two choices. In order to explore this aspect, the 20 teacher texts and the 380 student texts are successively modelled into k = 4 to k = 8 topics and the results are comparatively analyzed. Results are illustrated in Fig. 12 to Fig. 16, where teacher and student results are juxtaposed. Circles in these Figures correspond to questions within assignments. The LDA-derived topics are color-coded as in the respective Figure legends. Figure 12.a ( k = 4 , teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to Q1.1, Q1.2, Q2.1, Q2.2, Q2.3 all fall under topic_1 (brown), answers to Q3.1, Q3.2, Q3.3 fall under topic_3 (yellow), and then answers to Q4.1, Q5.1, Q5.4, Q5.5, Q6.1 fall under topic_2 (grey) and answers to Q4.2, Q4.3, Q5.2, Q5.3, Q6.2, Q6.3, Q6.4 fall under topic_0 (blue). This LDA-derived grouping suggests that (a) assignments 1 and 2 are closely related and might merge into a single new assignment, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into two new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5, Q6.1 and one that collects questions Q4.2, Q4.3, Q5.2, Q5.3, Q6.2, Q6.3, Q6.4. This recommendation is useful in case the teacher contemplates the restructuring of the course into 4 units instead of the current 6 ones. Figure 13.a ( k = 5 , teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1, Q2.2, Q2.3 all fall under topic_2 (grey), answers to Q3.1, Q3.2, Q3.3 fall under topic_1 (brown), answers to Q4.1, Q5.1, Q5.4, Q5.5, Q6.1 fall under topic_3 (yellow), answers to Q4.2, Q4.3, Q5.2, Q5.3 fall under topic_4 (light blue) and Q6.2, Q6.3, Q6.4 fall under topic_0 (dark blue). This LDA-derived grouping suggests that (a) assignments 1 and 2 are closely related and might merge into a single new assignment, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into 3 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5, Q6.1, one that collects questions Q4.2, Q4.3, Q5.2, Q5.3, and a last one that collects Q6.2, Q6.3, Q6.4. This recommendation is useful in case the teacher contemplates the restructuring of the course into 5 units instead of the current 6 ones. This recommendation is fairly similar to the previous one of k = 4 . It exploits the extra available topic to break the last group Q4.2, Q4.3, Q5.2, Q5.3, Q6.2, Q6.3, Q6.4 into two separate groups, Q4.2, Q4.3, Q5.2, Q5.3 and Q6.2, Q6.3, Q6.4, while the rest remain unchanged. Figure 14.a ( k = 6 , teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1 fall under topic_2 (grey), answers to Q2.2, Q2.3, Q6.1 fall under topic_4 (green), Q3.1, Q3.2, Q3.3 fall under topic_4 (light blue), answers to Q4.1, Q5.1, Q5.4, Q5.5 fall under topic_0 (dark blue), answers to Q4.2, Q4.3, Q5.2, Q5.3 fall under topic_1 (brown) and Q6.2, Q6.3, Q6.4 fall under topic_3 (yellow). This LDA-derived grouping suggests that (a) assignments 1 might annex Q2.1 of assignment 2 and leave assignment 2 with Q2.2, Q2.3 and Q.6.1 annexed from assignment 6, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4 and 5 intermingle and might be grouped into 2 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5 and one that collects questions Q4.2, Q4.3, Q5.2, Q5.3, (d) assignment 6, minus Q.6.1, may form a distinct assignment that collects Q6.2, Q6.3, Q6.4. This recommendation keeps the structure to the current 6 units but recommends to the teacher a restructuring of the questions within the assignments for a more coherent result. This recommendation is fairly similar to the previous one of k = 5 . It exploits the extra available topic to break assignment 2 into Q2.1 that is attached to assignment 1 and Q2.2, Q2.3 that jointly with annexed Q6.1 form a new group, while the rest remain unchanged. Figure 15.a ( k = 7 , teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1 fall under topic_4 (light blue), answers to Q2.2, Q2.3, Q6.1 fall under topic_1 (brown), Q3.1, Q3.2, Q3.3 fall under topic_6 (dark blue), answers to Q4.1, Q5.1, Q5.4, Q5.5 fall under topic_2 (grey), answer to Q4.2 falls under topic_5 (green), answers to Q4.3, Q5.2, Q5.3 and Q6.2, Q6.3, Q6.4 fall under topic_3 (yellow). This LDA-derived grouping suggests that (a) assignments 1 might annex Q2.1 of assignment 2 and leave assignment 2 with Q2.2, Q2.3 and Q.6.1 annexed from assignment 6, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into 3 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5, one with question Q4.2 only, and one with questions Q4.3, Q5.2, Q5.3, and Q6.2, Q6.3, Q6.4, (d) no text falls under topic_0 (medium blue). This recommendation keeps the structure to the current 6 units but recommends to the teacher a different restructuring than the previous case of k = 6 . It merges the last two groups of the k = 6 case (Q4.2, Q4.2, Q5.2, Q5.3 with Q6.2, Q6.3, Q.6.4) into one bigger group and then wastes the spared topic for the single question Q4.2. In comparison, if the teacher contemplates a structure of k = 6 units, the previous ( k = 6 ) case is preferable. Figure 16.a ( k = 8 , teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1, Q2.2 fall under topic_4 (light blue), answers to Q2.3, Q4.1, Q5.1, Q5.4, Q6.1 fall under topic_2 (grey), Q3.1, Q3.2, Q3.3 fall under topic_7 (dark brown), answers to Q4.2, Q4.3, Q5.2, Q5.3 fall under topic_5 (green), answer to question Q5.5 falls under topic_0 (medium blue) and answers to Q6.2, Q6.3, Q6.4 fall under topic_3 (yellow). No texts fall under topic_1 (light brown) or topic_6 (dark blue). This LDA-derived grouping suggests that (a) assignments 1 and 2 are closely related and might merge into a single new group, minus Q2.3, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into 3 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q6.1 and annexes Q2.3, one that collects questions Q.4.2, Q4.3, Q5.2, Q5.3, and one that groups the last three assignment 6 questions Q6.2, Q6.3, Q6.4. This is another recommendation that keeps the structure to the current 6 units but recommends to the teacher a different restructuring than the previous case of k = 6 . It bears element of both k = 6 and k = 7 cases, but recommends a rather unexpected group of Q4.1, Q5.1, Q5.4, Q6.1 and Q2.3, mixing questions from 4 assignments. In comparison, if the teacher contemplates a structure of k = 6 units, the previous ( k = 6 ) case is preferable. Figures 12.b to 16.b (student texts clustered into k = 4 to k = 8 topics, respectively) exhibit an increasing entropy with k . The recommendations they produce are not identical to those of the respective teacher texts in Figs. 12.a to 16.a; yet, these are found to be close while certain elements are repeatedly observed in student and teacher results alike (affinity between assignments 1 and 2, isolation of assignments 3, intermingled assignments 4, 5 and 6). Suggestions for the teacher are discussed in the Discussion section below. For a more detailed visualization of all 380 student texts mapped into a reduced dimensionality space that translates affinities into topological closeness, the t-distributed Stochastic Neighbor Embedding (t-SNE) method in the ORANGE tool is employed. t-SNE performs dimensionality reduction through the Principal Component Analysis (PCA) algorithm ( https://orange3.readthedocs.io/projects/orange-visual-programming/en/latest/widgets/unsupervised/tsne.html ). Results are shown in Fig. 17, in a zoom-out or overview fashion, for 3 different choices of the number of principal components used in PCA. Individual texts are identified as circled labeled by the adopted nomenclature of xyz.txt ( x -th assignment, y -th question, z -th student). As the number of PCA components increases from 15 to 17 to 20, visualization produces clusters of a better separation, as it may be verified by inspection of Fig. 17.a, 17.b, and 17.c, respectively. A closer inspection by zooming into Fig. 17, however, is more interesting as (a) it identifies questions within assignments with considerable accuracy, as manually marked on Fig. 18 ; (b) it reproduces to a considerable degree the affinities between specific questions that LDA-based topic modelling has revealed (Figs. 12 to 16). For example, in Fig. 19 three such close affinities are illustrated and visually verified by zooming into Fig. 17.b (t-SNE with 17 PCA components) (a) between answers to Q4.5 and Q5.5, in agreement to results in all Figs. 12–15; (b) among answers to Q5.1, Q5.4 and Q5.5, in agreement to results in all Figs. 12–15; ( c ) among answers to Q4.1, Q4.2, Q4.3, and Q.5.2, Q5.3, in agreement to results in all Figs. 12–14 and Fig. 16. Thanks to the reduction of dimensionality by t-SNE, the projections on the 2D plane obtained, as illustrated in Figs. 17, 18 and 19, constitute a verification of LDA-derived topic modelling results for k values in the vicinity of k = 6. Implications of these results for the teacher are discussed in the Discussion section below. 4 Discussion 4.1 RQ1: Can topic modeling cluster a body of texts consisting of student answers in-to ‘meaningful’ clusters, i.e., into clusters aligned to the inherent internal struc-ture and entities of the specific body of texts? On the ground of results obtained in sections 3.2 and 3.3 , it may be argued that indeed topic modelling by LDA can successfully cluster student answer texts into meaningful topics, that are aligned to a considerable degree to the internal structure of the data set (here, 6 assignments at a first level and 20 questions at a second, more detailed level). More specifically, accuracy scores calculated from confusion matrices in Tables 3 .a, and 3.b, at the coarser assignments level, and in Tables 6 .a and 6.b, at the finer questions level, indicate that the LDA-derived topics are in satisfactory agreement to the respective underlying course entities. 4.2 RQ2: Can the set of top words in each topic, extracted through topic modeling of student answers, successfully ‘represent’ the subset of texts clustered under that topic, i.e., allow the synthesis of a name or title for the texts clustered under that topic that is meaningful in the educational context of the specific body of texts? Qualitative evaluation of LDA-derived topic models was employed to answer this RQ. The preliminary experiment in section 3.1 answered it to the positive at the course level, as the top words extracted from the whole body of student answers were found to be ‘fully relevant’ to the course theme and capable of providing a meaningful title for it (Table 2 and Fig. 5 ). The answer to this RQ is based on results in Tables 4 and 5 (section 3.2 , assignments level) and Tables 7 and 8 (section 3.3 , questions level). Relevance scores are high in all three experiments, ranging between 3 / 4 (‘closely relevant’) and 4 / 4 (‘fully relevant’). Another interesting observation is that in all these cases, when the units are known (given), results are only a little higher than the respective results when the units are unknown (identified by LDA). In Table 5 , e.g., in student texts average relevance is 3.33 / 4.00 for LDA-derived units as compared to 3.42 / 4.00 for given units while in teacher texts average relevance is 3.33 / 4.00 for LDA-derived units as compared to 3.48 / 4.00 for given units. This outcome is a strong argument for the suitability of LDA in the educational context of the data set. 4.3 RQ3: (If RQ2 is answered to the positive) Is it possible to evaluate the quality of student answers by contrasting the topics and top words extracted from them against those extracted from teacher-provided model answers? The answer to this RQ is based on the percentage of common words between peer student and teacher lists across the clustering experiments. It is clear that the percentage of common top words between peer lists of student texts and teacher texts is considerably high (70.00%) at the course level (Table 2 , section 3.1 ), and gradually drops to 31.70% (LDA-derived units) or 41.33% (given units) at the assignments level (section 3.2 ) and then to 28.60% (LDA-derived units) or 36.00% (given units) at the questions level (section 3.3 ). Given that student evaluation and grading is performed at this finer (questions) level, however, on the basis of the current results the teacher is discouraged to evaluate the quality of student answers on the basis of top word comparison. 4.4 RQ4: In what practical ways, other than those implied in the previous RQs, can topic modelling be exploited by the teacher to improve his/her teaching? The results of the third experiment (exploratory procedure) can form the basis for an answer to this RQ. These results may be summarized as follows: 1. Course structure into k = 6 units and therefore k = 6 assignments for student assessment is the optimal choice in the vicinity of 6 and should be retained. 2. Assignments 1 and 2 are closely related and might be merged into a single new assignment. 3. Assignment 3 is isolated and should remain that way. 4. Assignments 4, 5 and 6 have overlapping content. Their questions might be rearranged in different ways; the most meaningful way is to place Q4.1, Q5.1, Q5.4, Q5.5 under one group, Q4.2, Q4.3, Q5.2, Q5.3 into a second group and leave Q6.1, Q6.2, Q6.3, Q6.4 into the same group they already belong. This last result comes from jointly considering the suggestions from Figs. 12 to 16, the visualizations in Figs. 17 to 19 and the experience of the class instructors with the course content. Indeed, this last element is the catalyst for the decisions as to which of the suggested restructuring solutions emerging from LDA topic modelling should be adopted. In the educational context of the current data set, for example, (i) Unit 4 (assignment 4 and questions therein) corresponds to Adaptive Learning Systems while Unit 5 (assignment 5 and questions therein) corresponds to Learning Styles and Personalized Learning. It is evident that these are closely connected themes, both in concept and in vocabulary. It is therefore not surprising that LDA-derived topics propose various ways to reorganize them internally. (ii) Unit 1 (assignment 1 and questions therein) corresponds to the Taxonomies of Learning while Unit 2 (assignment 2 and questions therein) corresponds to Learning Outcomes across multiple Domains of Learning. Again, it is clear that these two are closely connected themes, both in concept and in vocabulary. It is therefore not surprising that LDA-derived topics propose that they be merged into a single Unit (assignment). If this recommendation is adopted, course units drop from 6 to 5. This is an acceptably coherent situation, as verified by results in Figs. 13.a and 13.b. Of course, reorganization of the assignments and the questions within them are only meaningful is connection to the corresponding restructuring of the whole course and learning contents into units. 5 Conclusions Educational text mining has been applied in this study in an experimental setup based on a real data set of students answers and teacher model answers produced in the context of a graduate course offered in Greek language. The performance of LDA algorithm on topic modelling and top word extraction was experimentally assessed through a series of tests answering relevant research questions. Quantitative and qualitative results on evaluation of the LDA-derived topic models (sections 3.1 , 3.2 , 3.3 ) indicate that LDA can indeed produce topics and top word lists that are meaningful, i.e., aligned to the themes of the course and of its internal units, at various levels of detail, and therefore interpretable within the educational context of the specific course and data set. Building on the ‘confidence’ of this assertion, the exploratory experiment performed in the last part of this study (section 3.4 ) proceeded to evaluate alternative models that group the body of texts into either fewer or more clusters. The resulting LDA-derived topic models have been found to be directly interpretable within the context of the course and are therefore proposed as valid alternatives to the current course structure. Consequently, they qualify as practically useful recommendations for the teacher seeking to reorganize a course in an optimal way. This issue may arise either when the aim is for a more condensed / shrunk form of the course (fewer internal units, e.g., when the course has to be offered in fewer weeks) or for a more expanded / detailed form (more internal units, e.g., when the course has to be offered along more weeks). In either of these cases, the teacher may benefit from the alternative LDA-derived topic models in order to make a data-driven decision and adopt a meaningful new structure, both as to the study material modules and as to the assessment modules. Two such alternative structures have been discussed in detail in the Discussion section for the course used in the current study, as an indication of the potential of LDA-derived recommendations along that direction. In conclusion, a spectrum of interesting possibilities arises from the exploratory use of the topic modelling technology on educational data sets. Their comparative evaluation and possible adoption are critically dependent, however, on the human-in-the-loop – in that case, the class teachers. This view has been verified in the present study, where quality results have been possible only by leveraging on the teachers’ expertise on the specific taught subject. Table 6 a Confusion matrix of classification of 380 student texts under the k = 20 LDA-derived topics 1 topic_5 2 topic_14 3 topic_7 4 topic_19 5 topic_11 6 topic_2 7 topic_17 8 topic_18 9 topic_8 10 topic_3 11 topic_0 12 topic_6 13 topic_15 14 topic_10 15 topic_9 16 topic_16 17 topic_13 18 topic_4 19 topic_12 20 topic_1 Q1.1 0 16 1 2 Q1.2 19 Q2.1 16 1 2 Q2.2 19 Q2.3 12 7 Q3.1 15 2 Q3.2 1 13 4 Q3.3 16 0 1 1 Q4.1 1 17 1 Q4.2 5 12 2 Q4.3 4 15 Q5.1 19 Q5.2 4 15 Q5.3 1 18 Q5.4 1 1 10 7 Q5.5 2 17 Q6.1 2 2 15 Q6.2 2 5 3 0 9 Q6.3 19 Q6.4 3 1 15 Accuracy 68.3% Table 6 b Confusion matrix of classification of 20 teacher texts (one per question) under the k = 20 LDA-derived topics 1 topic_4 2 topic_0 3 topic_1 4 topic_2 5 topic_3 6 topic_9 7 topic_5 8 topic_6 9 topic_11 10 topic_8 11 topic_10 12 topic_7 13 topic_13 14 topic_14 15 topic_12 16 topic_16 17 topic_17 18 topic_15 19 topic_18 20 topic_19 Q1.1 1 Q1.2 1 Q2.1 1 Q2.2 1 Q2.3 1 Q3.1 1 Q3.2 1 Q3.3 1 Q4.1 1 Q4.2 1 Q4.3 1 Q5.1 1 Q5.2 1 Q5.3 1 Q5.4 1 Q5.5 1 Q6.1 1 Q6.2 1 Q6.3 1 Q6.4 1 Accuracy 33.5% Declarations Author Contribution Conceptualization, A.C., M.R. and D.M.; methodology, M.R., D.M. and D.K.; software, A.C.; validation, A.C., M.R. and D.M.; resources, M.R. and D.M.; writing-original draft preparation, A.C.; writing-review and editing, M.R. and D.K.; supervision, M.R. and D.K. All authors have read and agreed to the published version of the manuscript. Data Availability All data (student essays in Greek) analyzed in the current study are readily available from the corresponding author on reasonable request. References Berry, M. W., & Kogan, J. E. (2010). Text Mining: Applications and Theory . Wiley. https://doi.org/10.1002/9780470689646 . Goodfellow, I., Bengio, Y., & Courville, A. Deep learning.2016;MIT Press. Bilro, R. G., Loureiro, S. M. C., & Souto, P. (2023). A systematic review of customer behavior in business-to-business markets and agenda for future research. Journal of Business and Industrial Marketing , 38 (13), 122–142. https://doi.org/10.1108/JBIM-07-2022-0313 . Luque, C., Luna, J., Luque, M., & Ventura, S. (2019). An advanced review on text mining in medicine. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery , 9 . https://doi.org/10.1002/widm.1302 . Nandwani, P., & Verma, R. (2021). A review on sentiment analysis and emotion detection from text. Social Network Analysis and Mining , 11 (1), 81. https://doi.org/10.1007/s13278-021-00776-6 . Nowacki, M., & Stasiak, A. (2023). Identification of Memorable Game Experiences: Analysis of Escape Rooms Visitors’ Experiences. Games and Culture , 1–23. https://doi.org/10.1177/15554120231190648 . Shi, D., Zhou, J., Wu, F., Wang, D., Yang, D., & Pan, Q. (2024). Characteristics of students’ learning behavior preferences – an analysis of self-commentary data based on the LDA model. Journal of Intelligent and Fuzzy Systems , 46 (2), 4495–4509. https://doi.org/10.3233/JIFS-232971 . Agrawal, R. (2011). Enriching education through data mining. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). ;6911,1–2: https://doi.org/10.1007/978-3-642-23780-5_1 . Mello, R. F., André, M., Pinheiro, A., Costa, E., & Romero, C. (2019). Text mining in education. WIREs Data Mining Knowledge Discovery , 9 . https://doi.org/10.1002/widm.1332 . Deerwester, S., Dumais, S. T., Furnas, G. W., Landauer, T. K., & Harshman, R. (1990). Indexing by latent semantic analysis. Journal of the American Society for Information Scienc , 41 (6), 391–407. https://doi.org/10.1002/(SICI)1097-4571(199009)41:63.0.CO;2-9 . Dumais, S. T. Latent semantic indexing (LSI): Trec-3 report. In Proceedings of the Text REtrieval Conference (TREC-3).1995; D. Harman (Ed.) (219–230). Hofmann, T. (1999). Probabilistic Latent Semantic Indexing. In Proceedings of the 22nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Berkeley, USA. Association for Computing Machinery, New York, NY, USA. ;50–57: https://doi.org/10.1145/312624.312649 . Blei, D. M., Andrew, Y. N., & Jordan, M. I. (2003). Latent Dirichlet allocation. Journal of Machine Learning Research , 3 , 993–1022. Elrahman, A. A., Andreu, M., & Abbott, T. (2010). Using text data mining techniques for understanding freestyle question answers in course evaluation forms. Research in Higher Education Journal , 9 , 12–23. Stupans, I., McGuren, T., & Babey, A. M. (2016). Student Evaluation of Teaching: A Study Exploring Student Rating Instrument Free-form Text Comments. Innovative Higher Education , 41 , 33–42. https://doi.org/10.1007/s10755-015-9328-5 . Atapattu, T., Falkner, K., & Falkner, N. (2017). A comprehensive text analysis of lecture slides to generate concept maps. Computers & Education , 115 , 96–113. http://dx.doi.org/10.1016/j.compedu.2017.08.001 . Chen, X., Zou, D., Cheng, G., & Xie, H. Detecting latent topics and trends in educational technologies over four decades using structural topic modeling: A retrospective of all volumes of Computers & Education. Computers & Education.2020;151,038-1055: https://doi.org/10.1016/j.compedu.2020.103855 . Takei, M., Porter, S. R., Umbach, P. D., & Nakano, J. (2022). Understanding Themes in Postsecondary Research Using Topic Modeling and Journal Abstracts. Research in Higher Education (Springer, online). : https://doi.org/10.1007/s11162-023-09761-8 . Griffiths, T. L., & Steyvers, M. (2004). Finding scientific topics. Proceedings of the National Academy of Sciences of the United States of America , 101 , 5228–5235. https://doi.org/10.1073/pnas.0307752101 . Chen, N. S., Kinshuk Wei, C. W., & Chen, H. J. (2008). Mining e-learning domain concept map from academic articles. Computers & Education , 50 , 1009–1021. https://doi.org/10.1016/j.compedu.2006.10.001 . Hujala, M., Knutas, A., Hynninen, T., & Arminen, H. (2020). Improving the quality of teaching by utilising written student feedback:A stream lined process. Computers & Education , 157 , 1039–1065. https://doi.org/10.1016/j.compedu.2020.103965 . Molavi, M., Tavakoli, M., & Kismihók, G. Extracting Topics from Open Educational Resources. In: Lecture Notes in Computer Science Alario-Hoyos C, Rodríguez-Triana MJ, Scheffel M, Arnedillo-Sánchez I, Dennerlein SM, editors. Addressing Global Challenges and Quality Education. Springer, Cham EC-TEL,12315;2020: https://doi.org/10.1007/978-3-030-57717-9_44 . Sakhovskiy, A., Tutubalina, E., Solovyev, V., & Solnyshkina, M. (2023). Topic Modeling as a Method of Educational Text Structuring. In Proceedings of the IEEE 13th International Conference on Developments in eSystems Engineering (DeSE), Virtual Conference.: https://doi.org/10.1109/DeSE51703.2020.9450232 . Onah, D., Pang, E., El-Haj, M. A., & Data-driven (2022). Latent Semantic Analysis for Automatic Text Summarization using LDA Topic Modelling. In: Proceedings of the 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan.; : https://doi.org/10.1109/bigdata55660.2022.10020259 . Hoda, N., Ahmad, N., & Mahmood, M. R. (2022). Students’ Satisfaction with Technology-Assisted Learning: An Empirical Analysis of Female University Students in Saudi Arabia Using Telecourse Evaluation Questionnaire. In: Saini HS, Singh RK, Tariq Beg M, Mulaveesala R, Mahmood MR (Eds.) Innovations in Electronics and Communication Engineering. Lecture Notes in Networks and Systems. Springer, Singapore. ;355: https://doi.org/10.1007/978-981-16-8512-5_51 . May, D., Morkos, B., Jackson, A., Hunsu, N. J., Ingalls, A., & Beyette, F. (2022). Rapid transition of traditionally hands-on labs to online instruction in engineering courses. European Journal of Engineering Education , 48 , 842–860. https://doi.org/10.1080/03043797.2022.2046707 . Amashi, R., Dandgall, S. S., & Vijayalakshmi, M. (2023). Feedback for Faculty on Student’s Asynchronous Learning based on Classification using Topic Modelling. In: Proceedings of the IEEE Delhi Section Flagship Conference (DELCON), 2nd Edition, Rajpura, India.: https://doi.org/10.1109/DELCON57910.2023.10127267 . Jones, K. (1972). A statistical interpretation of term specificity and its application in retrieval. Journal of Documentation , 28 , 11–21. Salton, G., Wong, A., & Yang, C. (1975). A vector space model for automatic indexing. Communications of the ACM , 18 , 613–620. Litman, D. (2016). Natural language processing for enhancing teaching and learning. In: Proceedings of the 30th AAAI conference on artificial intelligence, AAAI Press, Phoenix, Arizona, USA. : https://doi.org/10.1609/aaai.v30i1.9879 . Shaik, T., Tao, X., LiY, Dann, C., Mcdonald, J., Redmond, P., & Galligan, L. (2022). A Review of the Trends and Challenges in Adopting Natural Language Processing Methods for Education Feedback Analysis. Ieee Access : Practical Innovations, Open Solutions , 10 , 56720–56739. https://doi.org/10.1109/ACCESS.2022.3177752 . Gottipati, S., Shankararaman, V., & Lin, J. R. (2018). Text analytics approach to extract course improvement suggestions from students’ feedback. Research and Practice in Technology Enhanced Learning , 13 . https://doi.org/10.1186/s41039-018-0073-0 . Zhang, L., Huang, Y., Yang, X., Yu, S., & Zhuang, F. (2022). An automatic short-answer grading model for semi-open-ended questions. Interactive Learning Environments , 30 , 177–190. https://doi.org/10.1080/10494820.2019.1648300 . Luo, W., & Litman, D. J. (2016). Determining the quality of a student reflective response. In: Proceedings of the 29th International Florida Artificial Intelligence Research Society Conference (FLAIRS), Key Largo, Florida, USA. ;226–231. El Kassas, W. S., Salama, C. R., Rafea, A. A., & Mohamed, H. K. Automatic text summarization: A comprehensive survey. Expert Systems with Applications .2021;165,113679: https://doi.org/10.1016/j.eswa.2020.113679 . Syed, S., & Spruit, M. Full-Text or Abstract? Examining Topic Coherence Scores Using Latent Dirichlet Allocation. In Proceedings of the IEEE International Conference on Data Science and Advanced Analytics (DSAA).2017;Tokyo, Japan (165–174): https://doi.org/10.1109/DSAA.2017.61 . Zhang, L., & VanLehn, K. (2017). Adaptively selecting biology questions generated from a semantic network. Interactive Learning Environments , 25 , 828–846. https://doi.org/10.1080/10494820.2016.1190939 . Kalamboukis, T. Z. (1995). Suffix stripping with modern Greek. Program , 29 , 313–321. Tambouratzis, G., & Carayannis, G. (2001). Automatic corpora-based stemming in Greek. Literacy and Linguistic Computing , 16 , 445–466. https://doi.org/10.1093/llc/16.4.445 . Ntais, G. (2006). Development of a stemmer for the Greek language. MSc Thesis, Stockholm University, Sweden. Saroukos, S. Enhancing a Greek Language Stemmer.2008;MSc Thesis, University of Tampere, Finland. Ntais, G., Saroukos, S., Berki, E., & Dalianis, H. (2016). Development and Enhancement of a Stemmer for the Greek Language. In: Proceedings of the 20th Pan-Hellenic Conference on Informatics (PCI '16), Association for Computing Machinery, New York, NY, USA.; : https://doi.org/10.1145/3003733.3003775 . Porter, M. F. (1980). An algorithm for suffix stripping. Program: automated library and information systems , 14 , 130–137. Pechlivanis, K., & Florou, E. (2015). : https://bitbucket.org/dataengineering/stemming/src/master/ , Accesed 25 January 2024. Pechlivanis, K., & Konstantopoulos, S. (2015). Corpus based methods for learning models of metaphor in modern Greek. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics. ; 9449:219–228: https://doi.org/10.1007/978-3-319-25789-1_21 . Rousseeuw, P. J., & Silhouettes (1987). A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics , 20 , 53–65. https://doi.org/10.1016/0377-0427(87)90125-7 . Davies, D., & Bouldin, D. (1979). A cluster separation measure. IEEE Transactions on Pattern Analysis and Machine Intelligence , 1 , 224–227. https://doi.org/10.1109/TPAMI.1979.4766909 . Calinski, T., & Harabasz, J. (1974). A dendrite method for cluster analysis. Communications in Statistics , 3 , 1–27. https://doi.org/10.1080/03610927408827101 . Hagen, L. (2018). Content analysis of e-petitions with topic modeling: How to train and evaluate LDA models? Information Processing and Management , 54 , 1292–1307. https://doi.org/10.1016/j.ipm.2018.05.006 . Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 17 Jan, 2025 Read the published version in Discover Computing → Version 1 posted Editorial decision: Revision requested 30 Jul, 2024 Reviews received at journal 29 Jul, 2024 Reviewers agreed at journal 29 Jul, 2024 Reviewers agreed at journal 10 Jul, 2024 Reviews received at journal 23 Jun, 2024 Reviewers agreed at journal 13 Jun, 2024 Reviewers invited by journal 12 Jun, 2024 Editor assigned by journal 04 Jun, 2024 Submission checks completed at journal 20 May, 2024 First submitted to journal 08 May, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4387141","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":304449265,"identity":"47fc447d-1416-4e45-a17a-176e7da0e5e6","order_by":0,"name":"Angelos Charitopoulos","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAx0lEQVRIie3NMQrCMBTG8Tc5BbqmS73Ck0BBkPYqLV3T4ujgkMluzh08hCDoagnoEvd4BxcRBKGKqaNL2k0kfwhZ3o8PwOX6yYh5COCZb9+HIPiiHzEz2PUevPJU32bTJmA6ryWZQ7QTFkJVkfkKkYW6SCQ5QLayzmmCvkBMt5qj5APIqE0MNWGPlmyqlrw6ENQk/KysqSH5AiIrGSkejgUyRtUF5XNJEysJjoqdRRMEXsnZtbpPYiv5jqaiL4G4t3C5XK6/7w04YzyHT7+I/QAAAABJRU5ErkJggg==","orcid":"","institution":"University of West Attica","correspondingAuthor":true,"prefix":"","firstName":"Angelos","middleName":"","lastName":"Charitopoulos","suffix":""},{"id":304449266,"identity":"3db07f17-e124-4819-8742-42a73aa2e882","order_by":1,"name":"Maria Rangoussi","email":"","orcid":"","institution":"University of West Attica","correspondingAuthor":false,"prefix":"","firstName":"Maria","middleName":"","lastName":"Rangoussi","suffix":""},{"id":304449267,"identity":"0dd95371-4bf0-46ad-9c5a-436cfea585a3","order_by":2,"name":"Dimitris Metafas","email":"","orcid":"","institution":"University of West Attica","correspondingAuthor":false,"prefix":"","firstName":"Dimitris","middleName":"","lastName":"Metafas","suffix":""},{"id":304449268,"identity":"e7cf5245-03d9-4b94-8eed-edefc4e8d155","order_by":3,"name":"Dimitrios Koulouriotis","email":"","orcid":"","institution":"National Technical University of Athens","correspondingAuthor":false,"prefix":"","firstName":"Dimitrios","middleName":"","lastName":"Koulouriotis","suffix":""}],"badges":[],"createdAt":"2024-05-08 06:59:35","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4387141/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4387141/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s10791-024-09496-9","type":"published","date":"2025-01-17T15:57:04+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":57482696,"identity":"5fe9b9b2-f1b5-4765-9482-3e068509f34a","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":78443,"visible":true,"origin":"","legend":"\u003cp\u003eData preprocessing in steps, as a RapidMiner two-level process - 1\u003csup\u003est\u003c/sup\u003e level: [process documents from files, write csv], [execute python], [read excel, nominal to text conversion, process documents from data]\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/37da314edba257263d010e42.png"},{"id":57483694,"identity":"fb246a84-f0fd-46fc-9d52-3618dd37db9e","added_by":"auto","created_at":"2024-05-31 09:40:50","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":76682,"visible":true,"origin":"","legend":"\u003cp\u003eData preprocessing in steps, as a RapidMiner two-level process - 2nd level: the sequence of 4 filters implemented within the process documents from data operator in Fig. 1\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/986de3f275c5258d34dfa765.png"},{"id":57482699,"identity":"9a279c7b-cb3a-4eb1-a0c7-3a6bdda54cb1","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":190813,"visible":true,"origin":"","legend":"\u003cp\u003eGreek stemmer: An excerpt of the Python script executed in RapidMiner\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/1283a7d939c8af2fe699aec6.png"},{"id":57483695,"identity":"f5f19ce5-983a-494e-a1dc-918befc28dd3","added_by":"auto","created_at":"2024-05-31 09:40:50","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":55190,"visible":true,"origin":"","legend":"\u003cp\u003eThe main processing step in RapidMiner: LDA loops over the designated set of preprocessed files\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/2d1b730b17c1acb8628b674e.png"},{"id":57483243,"identity":"209198ed-646f-4e4a-9e11-da8e2ac3eaca","added_by":"auto","created_at":"2024-05-31 09:32:50","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":69778,"visible":true,"origin":"","legend":"\u003cp\u003eTop 10 words in the total body of student texts (380) and teacher texts (20) and normalized weights to 1.00 (horizontal axis). Words are in Greek (stemmed). An English translation is provided in brackets, for readers’ convenience\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/133f5040e6e953d4cd8de5cc.png"},{"id":57483245,"identity":"3931dedd-5725-482d-876e-b7a693353a81","added_by":"auto","created_at":"2024-05-31 09:32:50","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":45636,"visible":true,"origin":"","legend":"\u003cp\u003eFirst clustering experiment: LDA-derived top 10 words within \u003cem\u003eLDA-extracted units\u003c/em\u003e (6 topics) – left frames, dashed lines\u003cem\u003e \u003c/em\u003eversus LDA-extracted top 10 words within\u003cem\u003e given units\u003c/em\u003e (6 assignments) – right frames, continuous lines. \u003cstrong\u003e(a)\u003c/strong\u003e Body of student texts (380) \u003cstrong\u003e(b)\u003c/strong\u003e Body of teacher texts (20)\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/514fd434095acfdc36ea1e50.png"},{"id":57483249,"identity":"3908ca0c-a383-449e-a25c-52c66613f9d6","added_by":"auto","created_at":"2024-05-31 09:32:51","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":91442,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e Visualization of the 6 x 6 confusion matrix in Table 3.a, for student texts clustered under k = 6 topics by LDA. X-axis: LDA-derived topics 1 to 6. Y-axis: percentage of texts belonging to a certain assignment (color-coded) that are clustered under the specific LDA-derived topic in the X-axis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb \u003c/strong\u003eVisualization of the 6 x 6 confusion matrix in Table 3.b, for teacher texts clustered under k = 6 topics by LDA. X-axis: LDA-derived topics 1 to 6. Y-axis: percentage of texts belonging to a certain assignment (color-coded) that are clustered under the specific LDA-derived topic in the X-axis\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/bc27ec926c3d8ec23bfc6a4d.png"},{"id":57482703,"identity":"10491564-744c-4a14-af80-3a1634c76d43","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":76150,"visible":true,"origin":"","legend":"\u003cp\u003eTop 10 words in Topic_2 (manually matched to Assignment 2) in student and teacher texts, and normalized weights to 1.00 (horizontal axis). Words are in Greek (stemmed). An English translation is provided in brackets, for readers’ convenience\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/3399687746b0bdd9bbab68c2.png"},{"id":57482711,"identity":"c2b62da7-b5d9-436a-9895-ca4a0fa2ab51","added_by":"auto","created_at":"2024-05-31 09:24:51","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":57114,"visible":true,"origin":"","legend":"\u003cp\u003eSecond clustering experiment: LDA-derived top 10 words within \u003cem\u003eLDA-extracted units\u003c/em\u003e (20 topics – left frames, dashed lines)\u003cem\u003e \u003c/em\u003eversus LDA-extracted top 10 words within\u003cem\u003e given units\u003c/em\u003e (20 questions – right frames, continuous lines). \u003cstrong\u003e(a)\u003c/strong\u003e Body of student texts (380) \u003cstrong\u003e(b)\u003c/strong\u003e Body of teacher texts (20)\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/d296a3a2402d8395bde1b979.png"},{"id":57482700,"identity":"4506ac3f-dfbe-497c-a9a7-3f8c4ac9383b","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":101259,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e Visualization of the confusion matrix for student texts classified under the k=20 topics by LDA. X-axis: LDA-derived topics 1 to 20. Y-axis: percentage of texts belonging to a certain assignment (color-coded) that are classified under a specific LDA-derived topic in the X-axis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e Visualization of the confusion matrix for teacher texts classified under the k=20 topics by LDA. X-axis: LDA-derived topics 1 to 20. Y-axis: percentage of texts belonging to a certain assignment (color-coded) that are classified under a specific LDA-derived topic in the X-axis\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/68704dd8d50cdc2ad61c1e50.png"},{"id":57482706,"identity":"a67efbf4-32ea-473f-b293-89b3b91ee9c9","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":22709,"visible":true,"origin":"","legend":"\u003cp\u003ePerplexity as a function of k, in the vicinity of k = 6\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/d91da062ba7f8d82707ded91.png"},{"id":57482704,"identity":"557129df-d54a-456d-a598-eec897f2079d","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":43019,"visible":true,"origin":"","legend":"\u003cp\u003eTeacher and student texts classified into k = 4 LDA-derived topics, color-coded as in the legend. X-axis: assignment nr. 1 to 6. Y-axis: nr. of texts (answers to questions) under the specific color-coded topic that belong to the specific assignment in the horizontal axis. (\u003cstrong\u003ea\u003c/strong\u003e) Teacher texts; (\u003cstrong\u003eb\u003c/strong\u003e) Student texts\u003c/p\u003e","description":"","filename":"12.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/9101b0748792fc1cd3b44c6e.png"},{"id":57483248,"identity":"9e9c37ae-08e0-46d5-8d1b-bd966ac55c6c","added_by":"auto","created_at":"2024-05-31 09:32:51","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":45452,"visible":true,"origin":"","legend":"\u003cp\u003eTeacher and student texts classified into k = 5 LDA-derived topics, color-coded as in the legend. X-axis: assignment nr. 1 to 6. Y-axis: nr. of texts (answers to questions) under the specific color-coded topic that belong to the specific assignment in the horizontal axis. (\u003cstrong\u003ea\u003c/strong\u003e) Teacher texts; (\u003cstrong\u003eb\u003c/strong\u003e) Student texts\u003c/p\u003e","description":"","filename":"13.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/8a8901ce0eeb336c1558e7b3.png"},{"id":57483247,"identity":"37dde805-b279-49ff-a0be-da0226e4b47b","added_by":"auto","created_at":"2024-05-31 09:32:50","extension":"png","order_by":14,"title":"Figure 14","display":"","copyAsset":false,"role":"figure","size":44963,"visible":true,"origin":"","legend":"\u003cp\u003eTeacher and student texts classified into k = 6 LDA-derived topics, color-coded as in the legend. X-axis: assignment nr. 1 to 6. Y-axis: nr. of texts (answers to questions) under the specific color-coded topic that belong to the specific assignment in the horizontal axis. (\u003cstrong\u003ea\u003c/strong\u003e) Teacher texts; (\u003cstrong\u003eb\u003c/strong\u003e) Student texts\u003c/p\u003e","description":"","filename":"14.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/cea81df0b30c9e8995e7ba78.png"},{"id":57482707,"identity":"8de3865c-8af1-4792-8e48-f947cbb30c67","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":15,"title":"Figure 15","display":"","copyAsset":false,"role":"figure","size":47720,"visible":true,"origin":"","legend":"\u003cp\u003eTeacher and student texts classified into k = 7 LDA-derived topics, color-coded as in the legend. X-axis: assignment nr. 1 to 6. Y-axis: nr. of texts (answers to questions) under the specific color-coded topic that belong to the specific assignment in the horizontal axis. (\u003cstrong\u003ea\u003c/strong\u003e) Teacher texts; (\u003cstrong\u003eb\u003c/strong\u003e) Student texts\u003c/p\u003e","description":"","filename":"15.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/15f1c07288153aa1033571c5.png"},{"id":57482709,"identity":"0acbc458-d81d-4572-8b43-94b00d1a134c","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":16,"title":"Figure 16","display":"","copyAsset":false,"role":"figure","size":46785,"visible":true,"origin":"","legend":"\u003cp\u003eTeacher and student texts classified into k = 8 LDA-derived topics, color-coded as in the legend. X-axis: assignment nr. 1 to 6. Y-axis: nr. of texts (answers to questions) under the specific color-coded topic that belong to the specific assignment in the horizontal axis. (\u003cstrong\u003ea\u003c/strong\u003e) Teacher texts; (\u003cstrong\u003eb\u003c/strong\u003e) Student texts\u003c/p\u003e","description":"","filename":"16.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/3dbea34a1b024f72a22066b9.png"},{"id":57482713,"identity":"6a1538e9-2565-447d-9125-37f5b7458c0c","added_by":"auto","created_at":"2024-05-31 09:24:51","extension":"png","order_by":17,"title":"Figure 17","display":"","copyAsset":false,"role":"figure","size":199372,"visible":true,"origin":"","legend":"\u003cp\u003eVisualization of the student texts (380) by t-SNE in ORANGE (zoom-out view). (\u003cstrong\u003ea\u003c/strong\u003e) PCA with 15 components; (\u003cstrong\u003eb\u003c/strong\u003e) PCA with 17 components; (\u003cstrong\u003ec\u003c/strong\u003e) PCA with 20 components\u003c/p\u003e","description":"","filename":"17.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/89b622acf298628e116f44a9.png"},{"id":57482710,"identity":"fa663c8b-341c-405a-a534-c7c4e2a7945f","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":18,"title":"Figure 18","display":"","copyAsset":false,"role":"figure","size":297447,"visible":true,"origin":"","legend":"\u003cp\u003eVisualization of the student texts (380) by t-SNE in ORANGE (Figure 17.b, 17 PCA components): clusters of texts answering to specific questions within assignments are identified (red circles)\u003c/p\u003e","description":"","filename":"18.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/1ce41e114392f2cd46a13975.png"},{"id":57482708,"identity":"ebe53c3b-4399-45ec-af8a-7fff1ba808c6","added_by":"auto","created_at":"2024-05-31 09:24:50","extension":"png","order_by":19,"title":"Figure 19","display":"","copyAsset":false,"role":"figure","size":198208,"visible":true,"origin":"","legend":"\u003cp\u003eAffinities among groups of answers to specific questions, as revealed by zooming into the visualization of the student texts (380) by t-SNE in ORANGE (Figure 18). (\u003cstrong\u003ea\u003c/strong\u003e) Affinity between answers to Q4.5 and Q5.5; (\u003cstrong\u003eb\u003c/strong\u003e) Affinity among answers to Q5.1, Q5.4 and Q5.5; (\u003cstrong\u003ec\u003c/strong\u003e) Affinity among answers to Q4.1, to Q4.2, Q4.3, and to Q.5.2, Q5.3\u003c/p\u003e","description":"","filename":"19.png","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/edeb34ec149d9e7c5f14fee6.png"},{"id":74284495,"identity":"be3ca6e8-4961-469a-bc02-876b206bb1e8","added_by":"auto","created_at":"2025-01-20 16:07:57","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4031340,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4387141/v1/57e57fa1-029e-40af-8d43-986825e09b54.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eText mining technologies applied to free-text answers of students in e-assessment: an experimental study in Greek \u003c/p\u003e","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eText (Data) Mining, a branch of the wider Data Mining area, is a collective term for sophisticated computational methods that analyze unstructured data in the form of texts to extract abstract information such as meaning, sentiments or opinions. Text Mining (TM) is today considered as a mature and yet impressively growing field [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Building on more than half a century of research and development, with milestones such as the advent of World Wide Web, the search engines on it (Google), Machine Learning and Deep Learning algorithms [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], Natural Language Processing (NLP) and commercial chatbots, and the current Large Language Models (LLM) such as OpenAI\u0026rsquo;s ChatGPT 3 and 4, TM has permeated a wide spectrum of fields such as business, e.g. [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e], healthcare, e.g. [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], communication, e.g. [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], entertainment, e.g. [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] and education, e.g. [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe affordances of TM for education have early been realized, investigated and exploited to improve teaching and learning [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], gradually shaping Educational Text Mining (ETM) as a distinct research area. The abundance of educational data, produced in digital form by Virtual Learning Environments (VLE), Open Educational Resources (OER), forums/chats/blogs and online collaborative environments, etc., along with the current advanced machine/deep learning algorithms for natural language processing, text classification and topic modeling, have allowed ETM to complement and enhance results of Educational Data Mining and Learning Analytics with results of text analysis and understanding. The interested reader is addressed to [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] and references therein for a comprehensive review on ETM.\u003c/p\u003e \u003cp\u003eIn the context of TM, topic modeling has emerged as an attempt to extract hidden (latent) themes or topics from single texts or bodies of texts and to group texts under these topics, in an unsupervised way. To this end, topic modelling has followed two major paths, the statistical approach, via Latent Semantic Indexing [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e],[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] and subsequently Probabilistic Latent Semantic Indexing, [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] and the machine learning approach, via Latent Dirichlet Allocation [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. The emergence of topic modeling has promoted education research to a higher level of sophistication, abstraction and scope, regarding the kind of questions addressed, e.g., [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e],[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eSince its introduction in 2003, Latent Dirichlet Allocation (LDA) algorithm has rapidly found extensive use in various fields, e.g. [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. In the scope of ETM, LDA has already been successfully employed for mining of concept maps from academic articles [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], feedback for teachers on evaluation of teaching by students [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], topic extraction from OER [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], structuring of teaching texts [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e], text summarization [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], student satisfaction analysis [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e], student attitudes towards online laboratories [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e], feedback for teachers on student performance [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], and many others.\u003c/p\u003e \u003cp\u003eA common thread that connects ETM studies is their aim to improve teaching and learning in practice: we have to fully understand teaching and learning in order to make it better. Concrete results on this aspect is the ultimate goal, implying that intelligence extracted from educational texts through text mining / topic modeling should be \u003cem\u003eactionable\u003c/em\u003e intelligence. Along that line, the need for automated and streamlined analysis and understanding of the various types of texts produced in educational contexts is evident today, due to the accelerating speed and volume of text generation in digital form. This target is being continuously sought by NLP technology. From its early stages [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] to the current state-of-the-art [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], NLP has created the high-quality text preprocessing foundation upon which current text mining / topic modeling algorithms working at the processing level have proliferated. The path is not yet fully automated, however; certainly, a human-in-the-loop is still necessary, even critical, for getting results \u0026ndash; [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e], [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] among others.\u003c/p\u003e \u003cp\u003eAmong the various research directions of ETM, an important one is student assessment through tests or assignments with open-ended questions that require answers in the form of free text. Evaluation and grading of such assignments is one of the most demanding and resource-consuming teacher tasks. It is not surprising, therefore, that considerable research effort has been dedicated to automate (parts of) that process and to evaluate the accuracy and efficiency of such automation, given the significance of (i) fair grading, in the case of summative assessment, or (ii) constructive feedback, in the case of formative assessment, for student progress. The automated assessment of short-answer tests [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e] and the extraction of feedback from reflective student responses [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] are examples of problems where conspicuous results have been obtained through NLP and ETM. For longer texts, considerable progress is made in text summarization [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e], [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] and text comparison and alignment [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Closely related to assessment of student answers is the automated generation of test questions within given topics, as an aid for the teacher [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe current study focuses on student assessment on the basis of student answers to open-ended questions, in the form of longer texts (essays). In contrast to existing research works, however, the aim is not to automate evaluation and grading or question generation. Here, the interest is on aiding the teacher evaluate the internal structure of his/her teaching content on the subject, as reflected in the structure of the corresponding evaluation assignments and questions therein, and possibly re-structure material and assignments jointly, towards a new organization consisting of more distinct and more internally coherent units. This is not an issue of better course material sequencing; rather, it is an issue of recognizing themes and subunits in the course content to increase coherence. Topic modelling is therefore a suitable approach, given its capacity to extract hidden topics and cluster texts (student answers, in our case) under each topic.\u003c/p\u003e \u003cp\u003eAnother aim of this study is to comparatively evaluate student answers by contrasting them to model answers prepared by the teacher for class-level feedback purposes. Comparison is attempted at the domain of top words representing each topic identified within the body of student answer texts on the one hand and the body of teacher model texts on the other hand. Incongruent topics, as revealed by disjoint top word sets, would alert the teacher that assignment questions in their current form lack clarity and comprehensibility and should be rethought and rephrased or more radically changed.\u003c/p\u003e \u003cp\u003eIn accordance with the above research aims, the following research questions (RQs) are posed:\u003c/p\u003e \u003cp\u003e1. RQ1: Can topic modeling cluster a body of texts consisting of student answers into \u0026lsquo;meaningful\u0026rsquo; clusters, i.e., into clusters aligned to the inherent internal structure and entities of the specific body of texts?\u003c/p\u003e \u003cp\u003e2. RQ2: Can the set of top words in each topic, extracted through topic modeling of student answers, successfully \u0026lsquo;represent\u0026rsquo; the subset of texts clustered under that topic, i.e., allow the synthesis of a name or title for the texts clustered under that topic that is meaningful in the educational context of the specific body of texts?\u003c/p\u003e \u003cp\u003e3. RQ3: (If RQ2 is answered to the positive) Is it possible to evaluate the quality of student answers by contrasting topics and top words extracted from them against those extracted from teacher-provided model answers?\u003c/p\u003e \u003cp\u003e4. RQ4: In what practical ways, other than those implied in the previous RQs, can topic modelling be exploited by the teacher to improve his/her teaching?\u003c/p\u003e \u003cp\u003eIn order to answer these RQs, an experimental study is carried out as described in the following sections. Answers are meant primarily as feedback or recommendation for the teacher, to help him/her (i) realize and identify strong and weak ties and relations between units and sub-units of the course material, as reflected in the structure and contents of the corresponding assignment of each unit, and (ii) evaluate the clarity and comprehensibility of assignment questions and proceed to change them, where necessary. Furthermore, the teacher may restructure and reorganize material and assignments and re-evaluate the results (the new student answers) using the same topic modelling approach until he/she achieves a desirable level of performance. In that sense, the proposed approach is a semi-manual support tool that can be used iteratively for the improvement of teaching.\u003c/p\u003e"},{"header":"2 Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Methodology\u003c/h2\u003e \u003cp\u003eMethodologically the present study is a quasi-experiment rather than an experiment because the data set used for analysis was collected during an existing semester-long graduate course not designed specifically for this study: (i) the texts produced by students already enrolled in this course constituted the sample of the study (convenience sampling) while (ii) the course regulation requires that all enrolled students be offered / taught the course in the prescribed way (no control group). Data collection, preprocessing and analysis were performed in batch mode, off-line, after the completion of the semester, as detailed in the following paragraphs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Data Set Preparation\u003c/h2\u003e \u003cp\u003eThe graduate course \u0026lsquo;E-learning Systems and Distance Learning Technologies\u0026rsquo; of the Master Degree Program \u0026lsquo;ICT for Education\u0026rsquo; was used as the source of texts for the present study. The 2022-23 spring semester class consisted of 19 students whose progress was evaluated through 6 assignments spread across the semester, one assignment every other week. Final course grade was the average of the (personal) best 5 out of 6 individual assignment grades. Each assignment consisted of a different number of questions depending on the subject, as detailed in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; therefore, every student should have answered a total of 20 questions by the end of the course. E-assessment was carried out in the moodle e-learning platform, running on the departmental moodle server. There was no time limitation other than the due time for turning the answers in, which was uniformly set to 2 weeks from assignment time.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eNumber of questions per assignment\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAssignment 1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment 2\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAssignment 3\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAssignment 4\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment 5\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAssignment 6\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ.1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ.2.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQ.3.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eQ.4.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQ.5.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eQ.6.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ.1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ.2.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQ.3.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eQ.4.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQ.5.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eQ.6.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQ.2.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQ.3.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eQ.4.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQ.5.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eQ.6.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQ.5.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eQ.6.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQ.5.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFor each subject treated in class, students were subsequently directed to study certain learning content made available on moodle, and then answer the corresponding assignment questions in the form of free-text answers, also in moodle. Grading and personalized feedback per question were also provided in moodle by the class instructors (2nd and 3rd authors of the current study). Furthermore, the class instructors composed and uploaded in moodle one set of \u0026lsquo;model\u0026rsquo; answers to all questions of the current assignment, in order to provide feedback at a non-personalized, class level. This course organization has produced (19 students x 20 answers/student) + (1 instructor x 20 answers/instructor)\u0026thinsp;=\u0026thinsp;400 texts, in total. The texts were extracted from moodle as 400 distinct _.txt files and renamed according to a convenient nomenclature as xyz.txt, where x\u0026thinsp;=\u0026thinsp;1, ..., 6 denotes the assignment number, y\u0026thinsp;=\u0026thinsp;2 or 3 or 4 or 5 denotes the number of questions within the specific (x-th) assignment, and z\u0026thinsp;=\u0026thinsp;1, ..., 19 denotes the student ID. For example, 4.3.18.txt is the answer of student with ID 18 to the 3rd question within the 4th assignment. As the working language in this Master Degree Program is Greek, questions and answers were composed in Greek \u0026ndash; a fact that proved to be a challenge in the subsequent steps of data preprocessing and data analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Data Preprocessing\u003c/h2\u003e \u003cp\u003eRapidMiner is the environment used to implement the successive steps of data preprocessing and data analysis (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://altair.com/altair-rapidminer\u003c/span\u003e\u003cspan address=\"https://altair.com/altair-rapidminer\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e ). Out of the variety of alternative environments, RapidMiner is selected thanks to its well-structured and self-evident interface, the wide spectrum of implemented algorithms, as well as the availability of sufficient documentation and help, as a result of its widespread use (REF). Certain limitations of RapidMiner regarding visualization of the analysis results, however, have led to the search for a complementary tool specialized in visualization, such as Alteryx (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.alteryx.com/\u003c/span\u003e\u003cspan address=\"https://www.alteryx.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e ) or Orange (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://orangedatamining.com/docs/\u003c/span\u003e\u003cspan address=\"https://orangedatamining.com/docs/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e ). The later was selected because it can handle Greek language texts/words.\u003c/p\u003e \u003cp\u003eRegarding data preprocessing, the typical sequence of steps is implemented in RapidMiner: data organization in 400 separate files, one file per answer text, named as described in the previous section and pre-stored in a local directory; data import to the RapidMiner environment in the form of a collection of text files; tokenization (breaking of each text into words); transformation into uniform-case fonts (here, uppercase); removal of stop-words (considered to carry no useful information as to the topic of the text); word stemming (removal of suffixes to reduce all inflexed forms of a given word into a single form); word filtering according to length in characters. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrate the relevant sequence of steps as a RapidMiner two-level process.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIt should be noted here that, unfortunately, Greek is not among the languages currently supported by RapidMiner \u0026ndash; or by any other equivalent environment, come to that. Special care has therefore been given to the language-dependent steps of\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003estop-word removal: A custom-made dictionary of Greek stop-words was manually compiled in the form of a .\u003cem\u003etxt\u003c/em\u003e file and used as a filter in RapidMiner to remove stop-words.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003estemming: During the last three decades intensive research in the field of Natural Language Processing for the Greek language has produced several stemmers, the major ones being TZK [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e], AMP [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e] and the Ntais Stemmer in its original and enhanced forms [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e], [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e], based on the Porter stemmer algorithm [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. The Greek stemmer more recently developed by [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e] in the context of the \u003cem\u003eMetaphor Detection\u003c/em\u003e project of NCSR \u0026lsquo;Demokritos\u0026rsquo; (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://metaphor.iit.demokritos.gr/\u003c/span\u003e\u003cspan address=\"http://metaphor.iit.demokritos.gr/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e ) is employed here [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. A Python implementation is available online in GitHub (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/kpech21/Greek-Stemmer\u003c/span\u003e\u003cspan address=\"https://github.com/kpech21/Greek-Stemmer\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e ), under GNU General Public License. This code was inserted in RapidMiner and executed on an installation of Python 3.10, using the \u0026lsquo;Execute Python\u0026rsquo; operator, to carry out stemming per text file. An excerpt of the python script executed in RapidMiner is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Data Analysis\u003c/h2\u003e \u003cp\u003eTopic modeling through the LDA algorithm [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] is employed in order to answer the two research questions defined at the outset of this study. The essential element in LDA is the introduction of an intermediate layer of \u0026lsquo;hidden\u0026rsquo; (latent) variables (the topics) between the apparent entities at the higher level (texts) and those at the lower level (words). LDA is implemented in RapidMiner (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eA major issue in topic modeling, common in all unsupervised methods, is the decision on the optimal number of topics, \u003cem\u003ek\u003c/em\u003e, which is required as an input parameter by all relevant algorithms. The same holds true for clustering algorithms that operate in the domain of numerical or categorical data instead of text data. For the latter case, \u003cem\u003ek\u003c/em\u003e is selected through various alternative methods, such as the Silhouette index [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e], the Davies-Bouldin index [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e] or the Calinski-Harabasz index [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]. For the case of text data, the optimal \u003cem\u003ek\u003c/em\u003e is usually sought by minimizing the perplexity or maximizing the loglikelihood of the clustered data set, across a range of values of \u003cem\u003ek\u003c/em\u003e. Decisions that favor empirically set values of \u003cem\u003ek\u003c/em\u003e, lower than the optimal value obtained through optimization, when the latter is impractically or unreasonably high given the application, are also described, e.g., [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn the present case, however, \u003cem\u003ek\u003c/em\u003e is considered as a known parameter: depending on whether the texts in the current data set are clustered (i) as to the assignment they answer to (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e) or, (ii) as to the question within the assignment they answer to (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e), or (iii) as to the ID of the student who authored them (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;19\u003c/em\u003e). The optimization step is therefore not necessary, in either of these cases. Similar, application-driven rather than optimization-driven selections of \u003cem\u003ek\u003c/em\u003e, are described, e.g., by [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] or by [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. It does remain meaningful, however, to perform a search in the vicinity of those initial values of \u003cem\u003ek\u003c/em\u003e, for values that represent a \u0026lsquo;better\u0026rsquo; or more \u0026lsquo;meaningful\u0026rsquo; clustering of the texts, and then try to interpret this \u0026lsquo;better\u0026rsquo; clustering and discuss its implications for the teacher.\u003c/p\u003e \u003cp\u003eConsequently, experimentation is organized into (i) a preliminary experiment, (ii) two major clustering experiments where k is known (given), and (iii) a final, exploratory experiment where a range of k values are investigated. In the following paragraphs, experiments are described and results are reported jointly for each part. A common thread across all experiments is the problem of content-based clustering of a body of texts, where \u0026lsquo;content\u0026rsquo; is represented by the topics extracted by LDA and the top words identified in each topic. The capacity of LDA to discern the internal structure inherent in a body of texts and to align the extracted topics to the native entities of this structure is investigated experimentally on the current data set. The (known) internal structure of the data set into assignments and further into questions facilitates the evaluation of the results and consequently the validation of LDA-derived topic models.\u003c/p\u003e \u003c/div\u003e"},{"header":"3 Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Preliminary experiment \u0026ndash; total body of texts\u003c/h2\u003e \u003cp\u003eThe preliminary experiment aims for a first, rough estimation of the capacity of LDA-based topic modelling to accurately represent the overall course contents. To this end, all 380 student texts are taken in a single cluster (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;1\u003c/em\u003e) and fed to LDA in order to extract the top 10 words to characterize it. For comparison purposes, the 20 teacher texts, again as a single cluster (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;1\u003c/em\u003e) are also fed to LDA for the extraction of the corresponding top 10 words. Results are given in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and visualized in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, where words are in Greek, stemmed and accompanied by an English translation in brackets, for readers\u0026rsquo; convenience. A qualitative evaluation of these results is performed by the 2 class instructors who, in the role of field experts, independently reviewed these two lists of top 10 words and scored their relevance to the theme of the course, on a 5-level Likert scale of relevance: {0\u0026thinsp;=\u0026thinsp;irrelevant; 1\u0026thinsp;=\u0026thinsp;loosely relevant; 2\u0026thinsp;=\u0026thinsp;moderately relevant; 3\u0026thinsp;=\u0026thinsp;closely relevant; 4\u0026thinsp;=\u0026thinsp;fully relevant or identical}. Final relevance scores, also shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e (bottom line), are calculated as the average of the two independent scores, where they do not differ by more than 2 levels in the scale; cases where independent scores differ by 3 levels or more are discussed and unanimously agreed scores are given.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTop 10 words in the total body of student texts (380) and teacher texts (20), and relevance scores\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIn Student Texts (380)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eweights\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003enormalized\u003c/p\u003e \u003cp\u003eweights\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIn Teacher Texts (20)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eweights\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003enormalized\u003c/p\u003e \u003cp\u003eweights\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e567\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e475\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e466\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e348\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e339\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e333\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e312\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e299\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.59\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eτροπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e279\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e263\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRELEVANCE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRELEVANCE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003csup\u003ea\u003c/sup\u003e shaded entries: common top words between student answers and teacher model answers.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAs it can be observed in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e out of the 10 top words (70%) are common in student answers and teacher model answers although not in the same ranking. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e reveals the strong alignment of terms employed by the students and the teachers; the fact that they do not fully coincide is welcome as it means that students do not replicate expressions and vocabulary found in the study material provided by the teacher; rather, they compose their own texts. Relevance scores indicate \u0026lsquo;full relevance\u0026rsquo; (4 / 4) both for the student top words and for the teacher top words. In fact, either of the two lists could help a researcher compose a title for this course \u0026lsquo;in blind\u0026rsquo;, having to do with \u0026ldquo;\u003cem\u003ethe effectiveness of (e-)learning-based educational systems regarding cognitive / learning gains of the learners\u003c/em\u003e\u0026rdquo; \u0026ndash; which is a satisfactory summary of the actual course content.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Clustering of Texts Through LDA \u0026ndash; Assignments Level\u003c/h2\u003e \u003cp\u003eThis first clustering experiment views the data set as a collection of 6 distinct subsets of texts, each subset consisting of all answers to the questions in a single assignment. Clustering of the 380 student texts under \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e LDA-derived topics is intended to investigate the potential of LDA to align the extracted topics to the inherent internal structure of the total body of texts, i.e., the assignments it consists of. The main processing step (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) is therefore executed once, with \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e, LDA parameters \u003cem\u003ea\u003c/em\u003e and \u003cem\u003eb\u003c/em\u003e empirically set to \u003cem\u003ea\u0026thinsp;=\u0026thinsp;0.1\u003c/em\u003e and \u003cem\u003eb\u0026thinsp;=\u0026thinsp;0.01\u003c/em\u003e and 2,500 iterations.\u003c/p\u003e \u003cp\u003eIdeally, each LDA-derived topic should uniquely identify with one of the 6 assignments, i.e., LDA should be able to (i) cluster under each of the identified \u003cem\u003ek\u003c/em\u003e topics those and only those student texts that answer questions in a single assignment; (ii) extract \u0026lsquo;meaningful\u0026rsquo; top words in each topic that accurately represent the theme assessed by the corresponding assignment. Accordingly, LDA performance in these two tasks is evaluated\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eregarding clustering: quantitatively, through accuracy (%) of the classification task, calculated as the percentage of the main diagonal values over the total values in the relevant 6 x 6 confusion matrix of topics v/s assignments;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eregarding top words: quantitatively, by the percentage of common top words between (a) the top 10 words in each LDA-extracted topic and (b) the top 10 words in the corresponding assignment, as these are extracted by running LDA independently 6 times, each time on the subset of student texts known to belong to the \u003cem\u003ei\u003c/em\u003e-th assignment (\u003cem\u003ei\u0026thinsp;=\u0026thinsp;1, \u0026hellip;, 6\u003c/em\u003e);\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eregarding top words: qualitatively, by the 2 class instructors who, in the role of field experts, reviewed the top 10 words extracted by LDA in each topic and scored their relevance to the theme of the corresponding assignment, on the same 5-level Likert scale adopted for the preliminary experiment and following the same procedure (averaging). The same procedure of relevance scoring they performed on the top 10 words derived by LDA independently in each assignment, the latter considered as a \u0026lsquo;benchmark\u0026rsquo; for the former.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eFor comparison purposes, the 20 teacher model answers are also pooled in a single body of texts and fed to LDA for topic modeling into \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e topics and the extraction of top 10 words in each topic.\u003c/p\u003e \u003cp\u003eThe experimental setup of this first clustering experiment is clarified in Fig.\u0026nbsp;6, where continuous lines are used to frame known/given units (6 assignments), while dashed lines are used to frame LDA-extracted units (6 topics).\u003c/p\u003e \u003cp\u003eQuantitative evaluation results are given in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.a [3.b] for student [teacher] texts, and illustrated in the corresponding Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e.a [7.b].\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003ea\u003c/b\u003e Student texts: confusion matrix of 6 assignments versus k\u0026thinsp;=\u0026thinsp;6 LDA-derived topics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003cp\u003e(topic_5)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003cp\u003e(topic_4)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003cp\u003e(topic_1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003cp\u003e(topic_3)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003cp\u003e(topic_2)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e6\u003c/p\u003e \u003cp\u003e(topic_0)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e38 / 38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e57 / 57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e53 / 54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.85%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.15%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e56 / 57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e18 / 57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e63.16%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e31.58%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e38 / 95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e56 / 95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.05%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e40.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e58.95%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e18 / 76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e58 / 76\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e23.18%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e76.30%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e66.78%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e Teacher texts: confusion matrix of 6 assignments versus k\u0026thinsp;=\u0026thinsp;6 LDA-derived topics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003cp\u003e(topic_2)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003cp\u003e(topic_5)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003cp\u003e(topic_4)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003cp\u003e(topic_1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003cp\u003e(topic_0)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e6\u003c/p\u003e \u003cp\u003e(topic_3)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 / 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 / 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 / 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e33.33%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e66.67%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3 / 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2 / 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1 / 3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e66.67%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e33.33%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2 / 5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e3 / 5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e40.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e60.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAssignm.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 / 4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e3 / 4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25.00%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e75.00%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e78.06%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eResults indicate that content-based clustering of texts according to LDA-derived topics is both feasible and meaningful in the educational context of the current data set. Indeed, and despite the non-ideal accuracy calculated for clustering student texts in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.a (66.78%) or teacher texts in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.b (78.06%), these percentages are significant and not compatible with \u0026lsquo;randomness\u0026rsquo; in the results. On the other hand, the off-diagonal entries in either of the two confusion matrices indicate to the teacher possible rearrangements across assignments, for a more coherent overall structure, as discussed in the \u003cspan refid=\"Sec12\" class=\"InternalRef\"\u003eDiscussion\u003c/span\u003e section below.\u003c/p\u003e \u003cp\u003eRegarding the top 10 words extracted from each topic, these are shown in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e, in Greek, stemmed, and sorted in descending order of weights. Top words extracted from student texts are shown in the left-hand side of Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e while those extracted from teacher texts are shown in the right-hand side of Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e, for comparison. The columns labeled as \u0026lsquo;Topic_i\u0026rsquo; (\u003cem\u003ei\u0026thinsp;=\u0026thinsp;1 .. 6\u003c/em\u003e) present the top words extracted by LDA from each LDA-extracted topic (dash-lined frames in Fig.\u0026nbsp;6), while the columns labeled as \u0026lsquo;Assignment_i\u0026rsquo; (\u003cem\u003ei\u0026thinsp;=\u0026thinsp;1 .. 6\u003c/em\u003e) present the top words extracted by LDA from each given subset of texts belonging to this assignment (continuous-lined frames in Fig.\u0026nbsp;6). The latter, therefore, act as \u0026lsquo;benchmarks\u0026rsquo; for the former. The Topic_2 zone of Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e is visualized in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e, to compare LDA-derived results between students and teacher (\u0026lsquo;Topics\u0026rsquo; columns only).\u003c/p\u003e \u003cp\u003eEach of the 3rd and 6th columns give the number of common top words over 10, between the two preceding columns. These number are averaged across the 6 groups in the bottom line of Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e. For student texts, common top words between given groups of texts (assignments) and LDA-derived groups of texts (topics) range from 2 / 10 (Topic_1 versus Assignment_1) to 10 / 10 (Topic_3 versus Assignment_3) with an average of 5.83 / 10 or 58.3%. For teacher texts, the range is 6 / 10 (Topic_1 versus Assignment_1) to 9 / 10 (Topic_3 versus Assignment_3) with an average of 7.16 / 10 or 71.6%. The fact that both these similarity scores are above 50% is satisfactory. At the same time, this quantitative evaluation is in good agreement to the accuracy results obtained in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.a (66.78%) for student texts and in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.b (78.06%) for teacher texts, where again teacher text scores are higher than student text scores.\u003c/p\u003e \u003cp\u003eA result that is probably more informative as to the quality of the obtained topic models is the percentage of common words\u003c/p\u003e \u003cp\u003e\u0026bull; between student texts (\u0026lsquo;Topics\u0026rsquo; column) and teacher texts (\u0026lsquo;Topics\u0026rsquo; column): this ranges from 0 / 10 to 6 / 10 with an average of 3.17 / 10 or 31.70%.\u003c/p\u003e \u003cp\u003e\u0026bull; between student texts (\u0026lsquo;Assignments\u0026rsquo; column) and teacher texts (\u0026lsquo;Assignments\u0026rsquo; column): this ranges from 2 / 10 to 6 / 10 with an average of 4.33 / 10 or 43.33%.\u003c/p\u003e \u003cp\u003eThese results are satisfactory, although somewhat lower when the units are extracted by LDA (\u0026lsquo;Topics\u0026rsquo; columns) than when the units are given (\u0026lsquo;Assignments\u0026rsquo; columns).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eStudent texts (left) and teacher texts (right): top 10 words of the LDA-extracted \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e topics and of the 6 assignments \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eStudent Texts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCommon\u003c/p\u003e \u003cp\u003ewords\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eTeacher Texts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eCommon\u003c/p\u003e \u003cp\u003ewords\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment_1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment_1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eθεωρ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eτροπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eβελτιως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eγνωστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eβελτιως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eερευν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eαρχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eατο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eδιαδικας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eση\u0026micro;αντικ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπληροφορ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eαξιολογης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eαρχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eδιαστας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eδη\u0026micro;ιουργ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment_2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment_2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eδιαδικας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eθεωρ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσκεψ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπεριγραφ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαξιολογης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπληροφορ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eθεωρ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eδη\u0026micro;ιουργ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eαξιολογης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eδιαδικας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eδιαδικας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπρακτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eδεξιοτητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment_3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment_3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eιστοσελιδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eιστοσελιδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eαισθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eκαλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eχρω\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eχρω\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eκανον\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπεριεχο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπληροφορ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eπληροφορ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eιστοτοπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eαισθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eκανον\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσελιδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπεριεχο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eκανον\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσελιδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eκανον\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eκαλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eιστοτοπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eκαλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eκαλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπολλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπολλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eκει\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eκακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eκακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eπεριεχο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eχρω\u0026micro;ατ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eχρω\u0026micro;ατ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eση\u0026micro;αντικ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eση\u0026micro;αντικ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσελιδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσελιδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπεριεχο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eκει\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;εγαλ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπληροφορ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment_4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment_4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπροσαρ\u0026micro;οστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eτεχνολογ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπροσαρ\u0026micro;οστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eτεχνολογ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπροσαρ\u0026micro;οστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eπροσαρ\u0026micro;οστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eασκης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eασκης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eαπαντης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπροβλη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eδιαφορετ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eυποστηριξ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eλυς\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eπροβλη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπλοηγης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eευφυ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment_5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment_5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπροβλη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθητυπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαπαντης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eερευν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eατο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθητυπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eτεχνολογ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eτροπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευς\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eλαθ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπροσαρ\u0026micro;οστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεπιλυς\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eτροπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eλυς\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eτροπ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eατο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eαποτελεσ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπροσφερ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eδιαφορετ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεννοι\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssignment_6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignment_6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυναισθη\u0026micro;ατ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυναισθη\u0026micro;ατ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυναισθη\u0026micro;ατ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυναισθη\u0026micro;ατ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυναισθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυναισθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eυπολογιστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eτεχνολογ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eυπολογιστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eτεχνολογ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eκαταστας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eκαταστας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eκαταστας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυστη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκφρας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eκαταστας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαναγνωρις\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eσυνθες\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυναισθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eχρηστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eφων\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυνθες\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eανθρωπιν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκφρας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;ηνυ\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eληψ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;οντελ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eευνοικ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eαποφας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eAverage common words\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.83 / 10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eAverage common words\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7.16 / 10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003csup\u003ea\u003c/sup\u003e shaded entries: common top words between student answers and peer teacher answers.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003e summarizes all relevance scores for top word lists from student texts and from teacher texts, as given by the 2 class instructors in the adopted 5-level scale. The columns labeled as \u0026lsquo;Topics\u0026rsquo; and \u0026lsquo;Assignments\u0026rsquo; in Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003e are aligned with the corresponding columns in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e. These qualitative evaluation results indicate that LDA-derived topic models as represented by the respective top 10 words are in good alignment to the educational context of the specific data set. Indeed, relevance scores in Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003e are in the three higher levels of the 5-level scale (0 to 4), with an average relevance of 3.33 / 4.00 or \u0026lsquo;closely relevant\u0026rsquo;, for words in LDA-derived topics from student texts (2nd column) and in LDA-derived topics from teacher texts (4th column) alike. Moreover, average relevance scores when units (assignments) are given are also close: 3.42 / 4.00 for student texts \u0026ndash; 3rd column against 3.58 / 4.00 for teacher texts \u0026ndash; 5th column. The facts that (i) individual entries as well as averages in adjacent columns under \u0026lsquo;Student Texts\u0026rsquo; are close, (ii) the same holds true for adjacent columns under \u0026lsquo;Teacher Texts\u0026rsquo;, while (iii) neither of the lists has any \u0026lsquo;irrelevant\u0026rsquo; or \u0026lsquo;loosely relevant\u0026rsquo; cases, are welcome since any significant difference in relevance scores, or numerous \u0026lsquo;irrelevant\u0026rsquo; or \u0026lsquo;loosely relevant\u0026rsquo; cases, would question the validity of the models. Furthermore, the fact that average relevance scores when the units are LDA-extracted (topics, 2nd and 4th columns) are only a little lower than when units are given (assignments, 3rd and 5th columns) is a direct result of the good quality of LDA clustering, as summarized in Tables\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.a and 3.b. It can therefore be claimed that quantitative and qualitative results are in good agreement.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRelevance scores for top 10 words in student texts and in teacher texts, across the 6 LDA-derived topics and the 6 given assignments\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eStudent Texts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eTeacher Texts\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003enr.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTopics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAssignments\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAssignments\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4.00\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.58\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Clustering of Texts Through LDA \u0026ndash; Questions Level\u003c/h2\u003e \u003cp\u003eIn the 2nd clustering experiment, the data set is considered as a collection of 20 distinct subsets of texts, each subset consisting of all student answers to a single question (note that questions across all 6 assignments add up to 20). Clustering of the 380 student texts under \u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e LDA-derived topics is intended to investigate the potential of LDA to align the extracted topics to the inherent internal structure of the total body of texts, i.e., the questions it consists of. This is a finer level of analysis compared to the assignments level investigated in the 1st experiment. The main processing step (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) is therefore executed once, with \u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e, LDA parameters \u003cem\u003ea\u003c/em\u003e and \u003cem\u003eb\u003c/em\u003e empirically set to \u003cem\u003ea\u0026thinsp;=\u0026thinsp;0.1\u003c/em\u003e and \u003cem\u003eb\u0026thinsp;=\u0026thinsp;0.01\u003c/em\u003e and 2,500 iterations.\u003c/p\u003e \u003cp\u003eFor comparison, the 20 teacher model answers are also pooled in a single body of texts and fed to LDA for topic modeling into \u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e topics.\u003c/p\u003e \u003cp\u003eThe setup of this 2nd experiment is clarified in Fig.\u0026nbsp;9, where continuous lines are used to frame known/given units (questions) while dashed lines are used to frame units extracted by LDA (topics).\u003c/p\u003e \u003cp\u003ePerformance of LDA models in this second experiment is evaluated quantitatively through confusion matrices and accuracy (%) scores and qualitatively, through top word comparison and relevance scores, as in the first experiment.\u003c/p\u003e \u003cp\u003eQuantitative results are shown in Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e6\u003c/span\u003e.a [Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e6\u003c/span\u003e.b] and visualized in Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e10\u003c/span\u003e.a [Figure \u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e10\u003c/span\u003e.b] for the student [teacher] texts. Accuracy calculated from the confusion matrix in Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e6\u003c/span\u003e.a [Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e6\u003c/span\u003e.b] is 66.93% [33.50%] for the student [teacher] texts clustered into 20 LDA-derived topics. Accuracy of student texts (66.93%) is at the same level with the accuracy obtained in the 1st experiment, for clustering of the same texts into \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e topics intended to align LDA-derived topics to assignments. The fact that performance is sustained when the number of topics \u003cem\u003ek\u003c/em\u003e (and, therefore, the complexity of the task) is dramatically increased from 6 to 20, is a strong indication that the adopted approach and the LDA-derived topic models are both feasible and meaningful in the educational context of the current data set. On the other hand, accuracy of teacher texts drops to 33.50% for \u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e topics, as compared to the 78.06% obtained for \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e topics in the first experiment. Closer inspection of Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e10\u003c/span\u003e.b reveals that this is due to the fact that LDA optimally clusters teacher texts under 7 (practically, under 6) topics, despite the \u0026lsquo;availability\u0026rsquo; of more topics when \u003cem\u003ek\u003c/em\u003e is set to 20. It is interesting that results are practically unaltered (7 topics, accuracy 40.0%) when LDA is re-run with a lower parameter \u003cem\u003ea\u0026thinsp;=\u0026thinsp;0.01\u003c/em\u003e, favoring an increased number of topics. This result is a strong indication for the teacher on the question of whether to further break down the course content to more than 6 units (assignments) or not, as discussed in the \u003cspan refid=\"Sec12\" class=\"InternalRef\"\u003eDiscussion\u003c/span\u003e section below.\u003c/p\u003e \u003cp\u003eTop 10 word lists for the 20 topics are not listed in full for this 2nd experiment, as was the case for the 1st experiment (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e), because of their volume. Instead, Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e shows the corresponding information of Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e for a single characteristic example among the 20 LDA-derived topics, namely, Topic_14, manually matched to question Q1.1 for the student texts and Topic_4, manually matched to Q1.1 for the teacher texts. The 3rd and 6th columns give the number of common top words over 10, between the two preceding columns, which are 7 / 10 for the student list and 8 / 10 for the teacher list. All common top word counts are averaged across the 20 lists and the results are shown in the bottom line of Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e. For student texts, common top words between given groups of texts (questions) and LDA-derived groups of texts (topics) range from 0 / 10 (Topic_18 versus Q3.3) to 10 / 10 (Topic_19 versus Q2.2) with an average of 5.60 / 10 or 56.0%. For teacher texts, the range is 2 / 10 (Topic_7 versus Q5.1) to 8 / 10 (Topic_4 versus Q1.1) with an average of 6 / 10 or 60.0%. The fact that both these similarity scores are above 50% is satisfactory. At the same time, this quantitative evaluation is in good agreement to the accuracy results obtained in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.a (66.78%) for student texts and in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.b (78.06%) for teacher texts, where again teacher text scores are higher than student text scores.\u003c/p\u003e \u003cp\u003eA result that is probably more informative as to the quality of the obtained topic models is the percentage of common words\u003c/p\u003e \u003cp\u003e(i) between student texts (\u0026lsquo;Topics\u0026rsquo; column) and teacher texts (\u0026lsquo;Topics\u0026rsquo; column): this ranges from 0 / 10 to 5 / 10 with an average of 2.86 / 10 or 28.6%.\u003c/p\u003e \u003cp\u003e(ii) between student texts (\u0026lsquo;Questions\u0026rsquo; column) and teacher texts (\u0026lsquo;Questions\u0026rsquo; column): this ranges from 1 / 10 to 6 / 10 with an average of 3.60 / 10 or 36.0%.\u003c/p\u003e \u003cp\u003eThese results are lower than the corresponding ones in the first experiment (31.70% and 43.33%, respectively) \u0026ndash; a direct impact of the increased complexity of this task (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e instead of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e) on the results. As in the 1st experiment, results are somewhat higher when the units are given (\u0026lsquo;Questions\u0026rsquo; columns) than when the units are extracted by LDA (\u0026lsquo;Topics\u0026rsquo; columns), which is an expected outcome.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eStudent texts (left) and teacher texts (right): a single example of a top 10 words list out of the 20 similar lists, comparing top 10 words of LDA-extracted \u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e topics against top 10 words of the question manually matched to each topic \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eStudent Texts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCommon\u003c/p\u003e \u003cp\u003ewords\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eTeacher Texts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eCommon\u003c/p\u003e \u003cp\u003eWords\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTopic_14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuestion_1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopic_4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQuestion_1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026micro;αθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεκπαιδευτ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεκπαιδευο\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eαξιολογης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγνωστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eδιαδικας\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθητ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγνως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eαξιολογης\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eβελτιως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eβελτιως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eεπιπεδ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eαρχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eαρχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026micro;αθη\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eσυγκεκρι\u0026micro;εν\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eγνωστ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eοργανως\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eταξινο\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eπρογρα\u0026micro;\u0026micro;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026micro;αθησιακ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eστοχ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026hellip;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eAverage common words\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.60 / 10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eAverage common words\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e6.00 / 10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003csup\u003ea\u003c/sup\u003e shaded entries: common top words between student answers and peer teacher answers.\u003c/p\u003e \u003cp\u003eQualitative results are based on the relevance scores given by the 2 class instructors who assessed the relevance of each of the 20 top word lists obtained from LDA-derived topics to the theme of the question manually matched to the specific topic. Results are gathered in Table\u0026nbsp;\u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e8\u003c/span\u003e for student texts (left side) and teacher texts (right side).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab8\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRelevance scores for top 10 words in student texts and in teacher texts, across the 20 LDA-derived topics and the 20 given questions (ordered by Question numbers)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eStudent Texts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eTeacher Texts\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003enr.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTopics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQuestions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTopics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eQuestions\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1 (Q1.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2 (Q1.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3 (Q2.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4 (Q2.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5 (Q2.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6 (Q3.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7 (Q3.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8 (Q3.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9 (Q4.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10 (Q4.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e11 (Q4.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e12 (Q5.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e13 (Q5.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e14 (Q5.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e15 (Q5.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e16 (Q5.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e17 (Q6.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e18 (Q6.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e19 (Q6.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e20 (Q6.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.13\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eRelevance scores in Table\u0026nbsp;\u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e8\u003c/span\u003e are at the upper 3 levels of the (0\u0026ndash;4) scale, indicating the high quality of the 20 LDA-extracted topics and their good correspondence to the 20 questions in this course. In the student texts, average relevance scores range from 2.0 to 4.0 with a mean value of 3.15 / 4.00 or \u0026lsquo;closely relevant\u0026rsquo;, obtained for LDA-extracted units (\u0026lsquo;Topics\u0026rsquo; column) and for given units (\u0026lsquo;Questions\u0026rsquo; column) alike. None of the cases is marked as \u0026lsquo;irrelevant\u0026rsquo; or \u0026lsquo;loosely relevant\u0026rsquo; \u0026ndash; a welcome result, since lower relevance scores with numerous \u0026lsquo;irrelevant\u0026rsquo; or \u0026lsquo;loosely relevant\u0026rsquo; cases, would question the validity of the LDA-derived models.\u003c/p\u003e \u003cp\u003eIt is interesting and worth reporting that a comparison of the top 10 student words in Table\u0026nbsp;\u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e8\u003c/span\u003e to the corresponding top 10 teacher words has not been possible. This is a direct consequence of the relevant comment made in the quantitative results of the k\u0026thinsp;=\u0026thinsp;20 experiment (previous paragraph). There, it was detected that when LDA runs on the body of 20 teacher texts with \u003cem\u003ek\u0026thinsp;=\u0026thinsp;20\u003c/em\u003e, it identifies and \u0026lsquo;populates\u0026rsquo; with top words only 7 topics, while the rest up to 20 are not exploited and remain empty of top words (4th column). For those 7 topics extracted from teacher texts and populated with top 10 words (4th column), relevance scores are close to the corresponding ones when the units are given (5th column), while their average across all 7 lists is 2.86 (\u0026lsquo;closely relevant\u0026rsquo;) that is a little lower than the average across all 20 (given) units in the 5th column (3.13, \u0026lsquo;closely relevant\u0026rsquo;).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Clustering of Texts Through LDA \u0026ndash; Search and optimization in the vicinity of k\u0026thinsp;=\u0026thinsp;6 topics\u003c/h2\u003e \u003cp\u003eThis last, exploratory experiment investigates the robustness of the structure of the course for certain alternative numbers of internal units (\u003cem\u003ek\u003c/em\u003e values). In the 2nd clustering experiment, where \u003cem\u003ek\u003c/em\u003e was set to 20, the 20 teacher texts (one per question) remained clustered under 6 or 7 topics, when more clusters were available. This inspired us to perform a search in the vicinity of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e for values of \u003cem\u003ek\u003c/em\u003e that might result in more coherent structures.\u003c/p\u003e \u003cp\u003ePractically, the 1st clustering experiment with \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e is repeated here in the vicinity of 6, namely, for each \u003cem\u003ek\u003c/em\u003e in the set {4, 5, 6, 7, 8}. Calculated perplexity values, shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e11\u003c/span\u003e as a function of \u003cem\u003ek\u003c/em\u003e, verify that \u003cem\u003ek\u0026thinsp;=\u0026thinsp;5, 6\u003c/em\u003e, 7 represent lower perplexities than \u003cem\u003ek\u0026thinsp;=\u0026thinsp;4\u003c/em\u003e or \u003cem\u003ek\u0026thinsp;=\u0026thinsp;8\u003c/em\u003e, possibly leading to more coherent structures. Although the current structure of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e does not represent the local minimum in this range, the previous (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;5\u003c/em\u003e) and the next (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;7\u003c/em\u003e) values are fairly close and practically equivalent. This result suggests that it is worth assessing the plausibility of either of these two choices.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn order to explore this aspect, the 20 teacher texts and the 380 student texts are successively modelled into \u003cem\u003ek\u0026thinsp;=\u0026thinsp;4\u003c/em\u003e to \u003cem\u003ek\u0026thinsp;=\u0026thinsp;8\u003c/em\u003e topics and the results are comparatively analyzed. Results are illustrated in Fig.\u0026nbsp;12 to Fig.\u0026nbsp;16, where teacher and student results are juxtaposed. Circles in these Figures correspond to questions within assignments. The LDA-derived topics are color-coded as in the respective Figure legends.\u003c/p\u003e \u003cp\u003eFigure 12.a (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;4\u003c/em\u003e, teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to Q1.1, Q1.2, Q2.1, Q2.2, Q2.3 all fall under topic_1 (brown), answers to Q3.1, Q3.2, Q3.3 fall under topic_3 (yellow), and then answers to Q4.1, Q5.1, Q5.4, Q5.5, Q6.1 fall under topic_2 (grey) and answers to Q4.2, Q4.3, Q5.2, Q5.3, Q6.2, Q6.3, Q6.4 fall under topic_0 (blue). This LDA-derived grouping suggests that (a) assignments 1 and 2 are closely related and might merge into a single new assignment, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into two new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5, Q6.1 and one that collects questions Q4.2, Q4.3, Q5.2, Q5.3, Q6.2, Q6.3, Q6.4. This recommendation is useful in case the teacher contemplates the restructuring of the course into 4 units instead of the current 6 ones.\u003c/p\u003e\u003cp\u003eFigure 13.a (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;5\u003c/em\u003e, teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1, Q2.2, Q2.3 all fall under topic_2 (grey), answers to Q3.1, Q3.2, Q3.3 fall under topic_1 (brown), answers to Q4.1, Q5.1, Q5.4, Q5.5, Q6.1 fall under topic_3 (yellow), answers to Q4.2, Q4.3, Q5.2, Q5.3 fall under topic_4 (light blue) and Q6.2, Q6.3, Q6.4 fall under topic_0 (dark blue). This LDA-derived grouping suggests that (a) assignments 1 and 2 are closely related and might merge into a single new assignment, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into 3 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5, Q6.1, one that collects questions Q4.2, Q4.3, Q5.2, Q5.3, and a last one that collects Q6.2, Q6.3, Q6.4. This recommendation is useful in case the teacher contemplates the restructuring of the course into 5 units instead of the current 6 ones. This recommendation is fairly similar to the previous one of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;4\u003c/em\u003e. It exploits the extra available topic to break the last group Q4.2, Q4.3, Q5.2, Q5.3, Q6.2, Q6.3, Q6.4 into two separate groups, Q4.2, Q4.3, Q5.2, Q5.3 and Q6.2, Q6.3, Q6.4, while the rest remain unchanged.\u003c/p\u003e \u003cp\u003eFigure 14.a (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e, teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1 fall under topic_2 (grey), answers to Q2.2, Q2.3, Q6.1 fall under topic_4 (green), Q3.1, Q3.2, Q3.3 fall under topic_4 (light blue), answers to Q4.1, Q5.1, Q5.4, Q5.5 fall under topic_0 (dark blue), answers to Q4.2, Q4.3, Q5.2, Q5.3 fall under topic_1 (brown) and Q6.2, Q6.3, Q6.4 fall under topic_3 (yellow). This LDA-derived grouping suggests that (a) assignments 1 might annex Q2.1 of assignment 2 and leave assignment 2 with Q2.2, Q2.3 and Q.6.1 annexed from assignment 6, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4 and 5 intermingle and might be grouped into 2 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5 and one that collects questions Q4.2, Q4.3, Q5.2, Q5.3, (d) assignment 6, minus Q.6.1, may form a distinct assignment that collects Q6.2, Q6.3, Q6.4. This recommendation keeps the structure to the current 6 units but recommends to the teacher a restructuring of the questions within the assignments for a more coherent result. This recommendation is fairly similar to the previous one of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;5\u003c/em\u003e. It exploits the extra available topic to break assignment 2 into Q2.1 that is attached to assignment 1 and Q2.2, Q2.3 that jointly with annexed Q6.1 form a new group, while the rest remain unchanged.\u003c/p\u003e \u003cp\u003eFigure 15.a (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;7\u003c/em\u003e, teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1 fall under topic_4 (light blue), answers to Q2.2, Q2.3, Q6.1 fall under topic_1 (brown), Q3.1, Q3.2, Q3.3 fall under topic_6 (dark blue), answers to Q4.1, Q5.1, Q5.4, Q5.5 fall under topic_2 (grey), answer to Q4.2 falls under topic_5 (green), answers to Q4.3, Q5.2, Q5.3 and Q6.2, Q6.3, Q6.4 fall under topic_3 (yellow). This LDA-derived grouping suggests that (a) assignments 1 might annex Q2.1 of assignment 2 and leave assignment 2 with Q2.2, Q2.3 and Q.6.1 annexed from assignment 6, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into 3 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q5.5, one with question Q4.2 only, and one with questions Q4.3, Q5.2, Q5.3, and Q6.2, Q6.3, Q6.4, (d) no text falls under topic_0 (medium blue). This recommendation keeps the structure to the current 6 units but recommends to the teacher a different restructuring than the previous case of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e. It merges the last two groups of the \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e case (Q4.2, Q4.2, Q5.2, Q5.3 with Q6.2, Q6.3, Q.6.4) into one bigger group and then wastes the spared topic for the single question Q4.2. In comparison, if the teacher contemplates a structure of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e units, the previous (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e) case is preferable.\u003c/p\u003e \u003cp\u003eFigure 16.a (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;8\u003c/em\u003e, teacher texts) reveals that, across the six assignments that span the horizontal axis, teacher answers to questions Q1.1, Q1.2, Q2.1, Q2.2 fall under topic_4 (light blue), answers to Q2.3, Q4.1, Q5.1, Q5.4, Q6.1 fall under topic_2 (grey), Q3.1, Q3.2, Q3.3 fall under topic_7 (dark brown), answers to Q4.2, Q4.3, Q5.2, Q5.3 fall under topic_5 (green), answer to question Q5.5 falls under topic_0 (medium blue) and answers to Q6.2, Q6.3, Q6.4 fall under topic_3 (yellow). No texts fall under topic_1 (light brown) or topic_6 (dark blue). This LDA-derived grouping suggests that (a) assignments 1 and 2 are closely related and might merge into a single new group, minus Q2.3, (b) assignment 3 is distinct and should remain as it is, (c) assignments 4, 5 and 6 intermingle and might be grouped into 3 new assignments, one that collects questions Q4.1, Q5.1, Q5.4, Q6.1 and annexes Q2.3, one that collects questions Q.4.2, Q4.3, Q5.2, Q5.3, and one that groups the last three assignment 6 questions Q6.2, Q6.3, Q6.4. This is another recommendation that keeps the structure to the current 6 units but recommends to the teacher a different restructuring than the previous case of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e. It bears element of both \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e and \u003cem\u003ek\u0026thinsp;=\u0026thinsp;7\u003c/em\u003e cases, but recommends a rather unexpected group of Q4.1, Q5.1, Q5.4, Q6.1 and Q2.3, mixing questions from 4 assignments. In comparison, if the teacher contemplates a structure of \u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e units, the previous (\u003cem\u003ek\u0026thinsp;=\u0026thinsp;6\u003c/em\u003e) case is preferable.\u003c/p\u003e \u003cp\u003eFigures 12.b to 16.b (student texts clustered into \u003cem\u003ek\u0026thinsp;=\u0026thinsp;4\u003c/em\u003e to \u003cem\u003ek\u0026thinsp;=\u0026thinsp;8\u003c/em\u003e topics, respectively) exhibit an increasing entropy with \u003cem\u003ek\u003c/em\u003e. The recommendations they produce are not identical to those of the respective teacher texts in Figs.\u0026nbsp;12.a to 16.a; yet, these are found to be close while certain elements are repeatedly observed in student and teacher results alike (affinity between assignments 1 and 2, isolation of assignments 3, intermingled assignments 4, 5 and 6). Suggestions for the teacher are discussed in the \u003cspan refid=\"Sec12\" class=\"InternalRef\"\u003eDiscussion\u003c/span\u003e section below.\u003c/p\u003e \u003cp\u003eFor a more detailed visualization of all 380 student texts mapped into a reduced dimensionality space that translates affinities into topological closeness, the t-distributed Stochastic Neighbor Embedding (t-SNE) method in the ORANGE tool is employed. t-SNE performs dimensionality reduction through the Principal Component Analysis (PCA) algorithm (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://orange3.readthedocs.io/projects/orange-visual-programming/en/latest/widgets/unsupervised/tsne.html\u003c/span\u003e\u003cspan address=\"https://orange3.readthedocs.io/projects/orange-visual-programming/en/latest/widgets/unsupervised/tsne.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e ). Results are shown in Fig.\u0026nbsp;17, in a zoom-out or overview fashion, for 3 different choices of the number of principal components used in PCA. Individual texts are identified as circled labeled by the adopted nomenclature of \u003cem\u003exyz.txt\u003c/em\u003e (\u003cem\u003ex\u003c/em\u003e-th assignment, \u003cem\u003ey\u003c/em\u003e-th question, \u003cem\u003ez\u003c/em\u003e-th student).\u003c/p\u003e \u003cp\u003eAs the number of PCA components increases from 15 to 17 to 20, visualization produces clusters of a better separation, as it may be verified by inspection of Fig.\u0026nbsp;17.a, 17.b, and 17.c, respectively. A closer inspection by zooming into Fig.\u0026nbsp;17, however, is more interesting as (a) it identifies questions within assignments with considerable accuracy, as manually marked on Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e18\u003c/span\u003e; (b) it reproduces to a considerable degree the affinities between specific questions that LDA-based topic modelling has revealed (Figs.\u0026nbsp;12 to 16). For example, in Fig.\u0026nbsp;19 three such close affinities are illustrated and visually verified by zooming into Fig.\u0026nbsp;17.b (t-SNE with 17 PCA components) (a) between answers to Q4.5 and Q5.5, in agreement to results in all Figs.\u0026nbsp;12\u0026ndash;15; (b) among answers to Q5.1, Q5.4 and Q5.5, in agreement to results in all Figs.\u0026nbsp;12\u0026ndash;15; (\u003cb\u003ec\u003c/b\u003e) among answers to Q4.1, Q4.2, Q4.3, and Q.5.2, Q5.3, in agreement to results in all Figs.\u0026nbsp;12\u0026ndash;14 and Fig.\u0026nbsp;16.\u003c/p\u003e \u003cp\u003eThanks to the reduction of dimensionality by t-SNE, the projections on the 2D plane obtained, as illustrated in Figs.\u0026nbsp;17, \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e18\u003c/span\u003e and 19, constitute a verification of LDA-derived topic modelling results for k values in the vicinity of k\u0026thinsp;=\u0026thinsp;6. Implications of these results for the teacher are discussed in the \u003cspan refid=\"Sec12\" class=\"InternalRef\"\u003eDiscussion\u003c/span\u003e section below.\u003c/p\u003e \u003c/div\u003e"},{"header":"4 Discussion","content":"\u003cp\u003e \u003cb\u003e4.1 RQ1: Can topic modeling cluster a body of texts consisting of student answers in-to \u0026lsquo;meaningful\u0026rsquo; clusters, i.e., into clusters aligned to the inherent internal struc-ture and entities of the specific body of texts?\u003c/b\u003e \u003c/p\u003e \u003cp\u003eOn the ground of results obtained in sections \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e3.2\u003c/span\u003e and \u003cspan refid=\"Sec10\" class=\"InternalRef\"\u003e3.3\u003c/span\u003e, it may be argued that indeed topic modelling by LDA can successfully cluster student answer texts into meaningful topics, that are aligned to a considerable degree to the internal structure of the data set (here, 6 assignments at a first level and 20 questions at a second, more detailed level). More specifically, accuracy scores calculated from confusion matrices in Tables\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.a, and 3.b, at the coarser assignments level, and in Tables\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e6\u003c/span\u003e.a and 6.b, at the finer questions level, indicate that the LDA-derived topics are in satisfactory agreement to the respective underlying course entities.\u003c/p\u003e \u003cp\u003e \u003cb\u003e4.2 RQ2: Can the set of top words in each topic, extracted through topic modeling of student answers, successfully \u0026lsquo;represent\u0026rsquo; the subset of texts clustered under that topic, i.e., allow the synthesis of a name or title for the texts clustered under that topic that is meaningful in the educational context of the specific body of texts?\u003c/b\u003e \u003c/p\u003e \u003cp\u003eQualitative evaluation of LDA-derived topic models was employed to answer this RQ. The preliminary experiment in section \u003cspan refid=\"Sec8\" class=\"InternalRef\"\u003e3.1\u003c/span\u003e answered it to the positive at the course level, as the top words extracted from the whole body of student answers were found to be \u0026lsquo;fully relevant\u0026rsquo; to the course theme and capable of providing a meaningful title for it (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). The answer to this RQ is based on results in Tables\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e4\u003c/span\u003e and \u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003e (section \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e3.2\u003c/span\u003e, assignments level) and Tables\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e and \u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e8\u003c/span\u003e (section \u003cspan refid=\"Sec10\" class=\"InternalRef\"\u003e3.3\u003c/span\u003e, questions level). Relevance scores are high in all three experiments, ranging between 3 / 4 (\u0026lsquo;closely relevant\u0026rsquo;) and 4 / 4 (\u0026lsquo;fully relevant\u0026rsquo;).\u003c/p\u003e \u003cp\u003eAnother interesting observation is that in all these cases, when the units are known (given), results are only a little higher than the respective results when the units are unknown (identified by LDA). In Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e5\u003c/span\u003e, e.g., in student texts average relevance is 3.33 / 4.00 for LDA-derived units as compared to 3.42 / 4.00 for given units while in teacher texts average relevance is 3.33 / 4.00 for LDA-derived units as compared to 3.48 / 4.00 for given units. This outcome is a strong argument for the suitability of LDA in the educational context of the data set.\u003c/p\u003e \u003cp\u003e \u003cb\u003e4.3 RQ3: (If RQ2 is answered to the positive) Is it possible to evaluate the quality of student answers by contrasting the topics and top words extracted from them against those extracted from teacher-provided model answers?\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe answer to this RQ is based on the percentage of common words between peer student and teacher lists across the clustering experiments. It is clear that the percentage of common top words between peer lists of student texts and teacher texts is considerably high (70.00%) at the course level (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, section \u003cspan refid=\"Sec8\" class=\"InternalRef\"\u003e3.1\u003c/span\u003e), and gradually drops to 31.70% (LDA-derived units) or 41.33% (given units) at the assignments level (section \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e3.2\u003c/span\u003e) and then to 28.60% (LDA-derived units) or 36.00% (given units) at the questions level (section \u003cspan refid=\"Sec10\" class=\"InternalRef\"\u003e3.3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eGiven that student evaluation and grading is performed at this finer (questions) level, however, on the basis of the current results the teacher is discouraged to evaluate the quality of student answers on the basis of top word comparison.\u003c/p\u003e \u003cp\u003e \u003cb\u003e4.4 RQ4: In what practical ways, other than those implied in the previous RQs, can topic modelling be exploited by the teacher to improve his/her teaching?\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe results of the third experiment (exploratory procedure) can form the basis for an answer to this RQ. These results may be summarized as follows:\u003c/p\u003e \u003cp\u003e1. Course structure into k\u0026thinsp;=\u0026thinsp;6 units and therefore k\u0026thinsp;=\u0026thinsp;6 assignments for student assessment is the optimal choice in the vicinity of 6 and should be retained.\u003c/p\u003e \u003cp\u003e2. Assignments 1 and 2 are closely related and might be merged into a single new assignment.\u003c/p\u003e \u003cp\u003e3. Assignment 3 is isolated and should remain that way.\u003c/p\u003e \u003cp\u003e4. Assignments 4, 5 and 6 have overlapping content. Their questions might be rearranged in different ways; the most meaningful way is to place Q4.1, Q5.1, Q5.4, Q5.5 under one group, Q4.2, Q4.3, Q5.2, Q5.3 into a second group and leave Q6.1, Q6.2, Q6.3, Q6.4 into the same group they already belong.\u003c/p\u003e \u003cp\u003eThis last result comes from jointly considering the suggestions from Figs.\u0026nbsp;12 to 16, the visualizations in Figs.\u0026nbsp;17 to 19 and the experience of the class instructors with the course content. Indeed, this last element is the catalyst for the decisions as to which of the suggested restructuring solutions emerging from LDA topic modelling should be adopted. In the educational context of the current data set, for example,\u003c/p\u003e \u003cp\u003e(i) Unit 4 (assignment 4 and questions therein) corresponds to Adaptive Learning Systems while Unit 5 (assignment 5 and questions therein) corresponds to Learning Styles and Personalized Learning. It is evident that these are closely connected themes, both in concept and in vocabulary. It is therefore not surprising that LDA-derived topics propose various ways to reorganize them internally.\u003c/p\u003e \u003cp\u003e(ii) Unit 1 (assignment 1 and questions therein) corresponds to the Taxonomies of Learning while Unit 2 (assignment 2 and questions therein) corresponds to Learning Outcomes across multiple Domains of Learning. Again, it is clear that these two are closely connected themes, both in concept and in vocabulary. It is therefore not surprising that LDA-derived topics propose that they be merged into a single Unit (assignment). If this recommendation is adopted, course units drop from 6 to 5. This is an acceptably coherent situation, as verified by results in Figs.\u0026nbsp;13.a and 13.b.\u003c/p\u003e \u003cp\u003eOf course, reorganization of the assignments and the questions within them are only meaningful is connection to the corresponding restructuring of the whole course and learning contents into units.\u003c/p\u003e"},{"header":"5 Conclusions","content":"\u003cp\u003eEducational text mining has been applied in this study in an experimental setup based on a real data set of students answers and teacher model answers produced in the context of a graduate course offered in Greek language. The performance of LDA algorithm on topic modelling and top word extraction was experimentally assessed through a series of tests answering relevant research questions. Quantitative and qualitative results on evaluation of the LDA-derived topic models (sections \u003cspan refid=\"Sec8\" class=\"InternalRef\"\u003e3.1\u003c/span\u003e, \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e3.2\u003c/span\u003e, \u003cspan refid=\"Sec10\" class=\"InternalRef\"\u003e3.3\u003c/span\u003e) indicate that LDA can indeed produce topics and top word lists that are meaningful, i.e., aligned to the themes of the course and of its internal units, at various levels of detail, and therefore interpretable within the educational context of the specific course and data set. Building on the \u0026lsquo;confidence\u0026rsquo; of this assertion, the exploratory experiment performed in the last part of this study (section \u003cspan refid=\"Sec11\" class=\"InternalRef\"\u003e3.4\u003c/span\u003e) proceeded to evaluate alternative models that group the body of texts into either fewer or more clusters. The resulting LDA-derived topic models have been found to be directly interpretable within the context of the course and are therefore proposed as valid alternatives to the current course structure. Consequently, they qualify as practically useful recommendations for the teacher seeking to reorganize a course in an optimal way. This issue may arise either when the aim is for a more condensed / shrunk form of the course (fewer internal units, e.g., when the course has to be offered in fewer weeks) or for a more expanded / detailed form (more internal units, e.g., when the course has to be offered along more weeks). In either of these cases, the teacher may benefit from the alternative LDA-derived topic models in order to make a data-driven decision and adopt a meaningful new structure, both as to the study material modules and as to the assessment modules. Two such alternative structures have been discussed in detail in the \u003cspan refid=\"Sec12\" class=\"InternalRef\"\u003eDiscussion\u003c/span\u003e section for the course used in the current study, as an indication of the potential of LDA-derived recommendations along that direction. In conclusion, a spectrum of interesting possibilities arises from the exploratory use of the topic modelling technology on educational data sets. Their comparative evaluation and possible adoption are critically dependent, however, on the human-in-the-loop \u0026ndash; in that case, the class teachers. This view has been verified in the present study, where quality results have been possible only by leveraging on the teachers\u0026rsquo; expertise on the specific taught subject.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab9\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003ea\u003c/b\u003e Confusion matrix of classification of 380 student texts under the k\u0026thinsp;=\u0026thinsp;20 LDA-derived topics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"21\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c14\" colnum=\"14\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c15\" colnum=\"15\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c16\" colnum=\"16\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c17\" colnum=\"17\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c18\" colnum=\"18\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c19\" colnum=\"19\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c20\" colnum=\"20\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c21\" colnum=\"21\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003cp\u003etopic_5\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003cp\u003etopic_14\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003cp\u003etopic_7\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003cp\u003etopic_19\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003cp\u003etopic_11\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e6\u003c/p\u003e \u003cp\u003etopic_2\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e7\u003c/p\u003e \u003cp\u003etopic_17\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003e8\u003c/p\u003e \u003cp\u003etopic_18\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003e9\u003c/p\u003e \u003cp\u003etopic_8\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003e10\u003c/p\u003e \u003cp\u003etopic_3\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c12\"\u003e \u003cp\u003e11\u003c/p\u003e \u003cp\u003etopic_0\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c13\"\u003e \u003cp\u003e12\u003c/p\u003e \u003cp\u003etopic_6\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c14\"\u003e \u003cp\u003e13\u003c/p\u003e \u003cp\u003etopic_15\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c15\"\u003e \u003cp\u003e14\u003c/p\u003e \u003cp\u003etopic_10\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c16\"\u003e \u003cp\u003e15\u003c/p\u003e \u003cp\u003etopic_9\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c17\"\u003e \u003cp\u003e16\u003c/p\u003e \u003cp\u003etopic_16\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c18\"\u003e \u003cp\u003e17\u003c/p\u003e \u003cp\u003etopic_13\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c19\"\u003e \u003cp\u003e18\u003c/p\u003e \u003cp\u003etopic_4\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c20\"\u003e \u003cp\u003e19\u003c/p\u003e \u003cp\u003etopic_12\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c21\"\u003e \u003cp\u003e20\u003c/p\u003e \u003cp\u003etopic_1\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c19\" namest=\"c18\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e \u003cp\u003e68.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab10\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003eb\u003c/b\u003e Confusion matrix of classification of 20 teacher texts (one per question) under the k\u0026thinsp;=\u0026thinsp;20 LDA-derived topics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"21\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c14\" colnum=\"14\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c15\" colnum=\"15\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c16\" colnum=\"16\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c17\" colnum=\"17\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c18\" colnum=\"18\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c19\" colnum=\"19\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c20\" colnum=\"20\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c21\" colnum=\"21\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003cp\u003etopic_4\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003cp\u003etopic_0\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003cp\u003etopic_1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003cp\u003etopic_2\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003cp\u003etopic_3\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e6\u003c/p\u003e \u003cp\u003etopic_9\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e7\u003c/p\u003e \u003cp\u003etopic_5\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003e8\u003c/p\u003e \u003cp\u003etopic_6\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003e9\u003c/p\u003e \u003cp\u003etopic_11\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003e10\u003c/p\u003e \u003cp\u003etopic_8\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c12\"\u003e \u003cp\u003e11\u003c/p\u003e \u003cp\u003etopic_10\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c13\"\u003e \u003cp\u003e12\u003c/p\u003e \u003cp\u003etopic_7\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c14\"\u003e \u003cp\u003e13\u003c/p\u003e \u003cp\u003etopic_13\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c15\"\u003e \u003cp\u003e14\u003c/p\u003e \u003cp\u003etopic_14\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c16\"\u003e \u003cp\u003e15\u003c/p\u003e \u003cp\u003etopic_12\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c17\"\u003e \u003cp\u003e16\u003c/p\u003e \u003cp\u003etopic_16\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c18\"\u003e \u003cp\u003e17\u003c/p\u003e \u003cp\u003etopic_17\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c19\"\u003e \u003cp\u003e18\u003c/p\u003e \u003cp\u003etopic_15\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c20\"\u003e \u003cp\u003e19\u003c/p\u003e \u003cp\u003etopic_18\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c21\"\u003e \u003cp\u003e20\u003c/p\u003e \u003cp\u003etopic_19\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ5.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ6.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c18\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c19\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c12\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c13\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c14\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c15\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c16\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c17\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c19\" namest=\"c18\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c20\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c21\"\u003e \u003cp\u003e33.5%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConceptualization, A.C., M.R. and D.M.; methodology, M.R., D.M. and D.K.; software, A.C.; validation, A.C., M.R. and D.M.; resources, M.R. and D.M.; writing-original draft preparation, A.C.; writing-review and editing, M.R. and D.K.; supervision, M.R. and D.K. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll data (student essays in Greek) analyzed in the current study are readily available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBerry, M. W., \u0026amp; Kogan, J. E. (2010). \u003cem\u003eText Mining: Applications and Theory\u003c/em\u003e. Wiley. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/9780470689646\u003c/span\u003e\u003cspan address=\"10.1002/9780470689646\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoodfellow, I., Bengio, Y., \u0026amp; Courville, A. Deep learning.2016;MIT Press.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBilro, R. G., Loureiro, S. M. C., \u0026amp; Souto, P. (2023). A systematic review of customer behavior in business-to-business markets and agenda for future research. \u003cem\u003eJournal of Business and Industrial Marketing\u003c/em\u003e, \u003cem\u003e38\u003c/em\u003e(13), 122\u0026ndash;142. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1108/JBIM-07-2022-0313\u003c/span\u003e\u003cspan address=\"10.1108/JBIM-07-2022-0313\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuque, C., Luna, J., Luque, M., \u0026amp; Ventura, S. (2019). An advanced review on text mining in medicine. \u003cem\u003eWiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/widm.1302\u003c/span\u003e\u003cspan address=\"10.1002/widm.1302\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNandwani, P., \u0026amp; Verma, R. (2021). A review on sentiment analysis and emotion detection from text. \u003cem\u003eSocial Network Analysis and Mining\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(1), 81. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s13278-021-00776-6\u003c/span\u003e\u003cspan address=\"10.1007/s13278-021-00776-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNowacki, M., \u0026amp; Stasiak, A. (2023). Identification of Memorable Game Experiences: Analysis of Escape Rooms Visitors\u0026rsquo; Experiences. \u003cem\u003eGames and Culture\u003c/em\u003e, 1\u0026ndash;23. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/15554120231190648\u003c/span\u003e\u003cspan address=\"10.1177/15554120231190648\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShi, D., Zhou, J., Wu, F., Wang, D., Yang, D., \u0026amp; Pan, Q. (2024). Characteristics of students\u0026rsquo; learning behavior preferences \u0026ndash; an analysis of self-commentary data based on the LDA model. \u003cem\u003eJournal of Intelligent and Fuzzy Systems\u003c/em\u003e, \u003cem\u003e46\u003c/em\u003e(2), 4495\u0026ndash;4509. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3233/JIFS-232971\u003c/span\u003e\u003cspan address=\"10.3233/JIFS-232971\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAgrawal, R. (2011). Enriching education through data mining. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). ;6911,1\u0026ndash;2: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-642-23780-5_1\u003c/span\u003e\u003cspan address=\"10.1007/978-3-642-23780-5_1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMello, R. F., Andr\u0026eacute;, M., Pinheiro, A., Costa, E., \u0026amp; Romero, C. (2019). Text mining in education. \u003cem\u003eWIREs Data Mining Knowledge Discovery\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/widm.1332\u003c/span\u003e\u003cspan address=\"10.1002/widm.1332\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeerwester, S., Dumais, S. T., Furnas, G. W., Landauer, T. K., \u0026amp; Harshman, R. (1990). Indexing by latent semantic analysis. \u003cem\u003eJournal of the American Society for Information Scienc\u003c/em\u003e, \u003cem\u003e41\u003c/em\u003e(6), 391\u0026ndash;407. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/(SICI)1097-4571(199009)41:6\u0026lt;391::AID-ASI1\u0026gt;3.0.CO;2-9\u003c/span\u003e\u003cspan address=\"10.1002/(SICI)1097-4571(199009)41:6%3C391::AID-ASI1%3E3.0.CO;2-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDumais, S. T. Latent semantic indexing (LSI): Trec-3 report. In Proceedings of the Text REtrieval Conference (TREC-3).1995; D. Harman (Ed.) (219\u0026ndash;230).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHofmann, T. (1999). Probabilistic Latent Semantic Indexing. In Proceedings of the 22nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Berkeley, USA. Association for Computing Machinery, New York, NY, USA. ;50\u0026ndash;57: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/312624.312649\u003c/span\u003e\u003cspan address=\"10.1145/312624.312649\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlei, D. M., Andrew, Y. N., \u0026amp; Jordan, M. I. (2003). Latent Dirichlet allocation. \u003cem\u003eJournal of Machine Learning Research\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e, 993\u0026ndash;1022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElrahman, A. A., Andreu, M., \u0026amp; Abbott, T. (2010). Using text data mining techniques for understanding freestyle question answers in course evaluation forms. \u003cem\u003eResearch in Higher Education Journal\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 12\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStupans, I., McGuren, T., \u0026amp; Babey, A. M. (2016). Student Evaluation of Teaching: A Study Exploring Student Rating Instrument Free-form Text Comments. \u003cem\u003eInnovative Higher Education\u003c/em\u003e, \u003cem\u003e41\u003c/em\u003e, 33\u0026ndash;42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10755-015-9328-5\u003c/span\u003e\u003cspan address=\"10.1007/s10755-015-9328-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAtapattu, T., Falkner, K., \u0026amp; Falkner, N. (2017). A comprehensive text analysis of lecture slides to generate concept maps. \u003cem\u003eComputers \u0026amp; Education\u003c/em\u003e, \u003cem\u003e115\u003c/em\u003e, 96\u0026ndash;113. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://dx.doi.org/10.1016/j.compedu.2017.08.001\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2017.08.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, X., Zou, D., Cheng, G., \u0026amp; Xie, H. Detecting latent topics and trends in educational technologies over four decades using structural topic modeling: A retrospective of all volumes of Computers \u0026amp; Education. Computers \u0026amp; Education.2020;151,038-1055: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2020.103855\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2020.103855\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTakei, M., Porter, S. R., Umbach, P. D., \u0026amp; Nakano, J. (2022). Understanding Themes in Postsecondary Research Using Topic Modeling and Journal Abstracts. Research in Higher Education (Springer, online). :\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11162-023-09761-8\u003c/span\u003e\u003cspan address=\"10.1007/s11162-023-09761-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGriffiths, T. L., \u0026amp; Steyvers, M. (2004). Finding scientific topics. \u003cem\u003eProceedings of the National Academy of Sciences of the United States of America\u003c/em\u003e, \u003cem\u003e101\u003c/em\u003e, 5228\u0026ndash;5235. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1073/pnas.0307752101\u003c/span\u003e\u003cspan address=\"10.1073/pnas.0307752101\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, N. S., Kinshuk Wei, C. W., \u0026amp; Chen, H. J. (2008). Mining e-learning domain concept map from academic articles. \u003cem\u003eComputers \u0026amp; Education\u003c/em\u003e, \u003cem\u003e50\u003c/em\u003e, 1009\u0026ndash;1021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2006.10.001\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2006.10.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHujala, M., Knutas, A., Hynninen, T., \u0026amp; Arminen, H. (2020). Improving the quality of teaching by utilising written student feedback:A stream lined process. \u003cem\u003eComputers \u0026amp; Education\u003c/em\u003e, \u003cem\u003e157\u003c/em\u003e, 1039\u0026ndash;1065. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2020.103965\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2020.103965\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMolavi, M., Tavakoli, M., \u0026amp; Kismih\u0026oacute;k, G. Extracting Topics from Open Educational Resources. In: Lecture Notes in Computer Science Alario-Hoyos C, Rodr\u0026iacute;guez-Triana MJ, Scheffel M, Arnedillo-S\u0026aacute;nchez I, Dennerlein SM, editors. Addressing Global Challenges and Quality Education. Springer, Cham EC-TEL,12315;2020: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-030-57717-9_44\u003c/span\u003e\u003cspan address=\"10.1007/978-3-030-57717-9_44\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSakhovskiy, A., Tutubalina, E., Solovyev, V., \u0026amp; Solnyshkina, M. (2023). Topic Modeling as a Method of Educational Text Structuring. In Proceedings of the IEEE 13th International Conference on Developments in eSystems Engineering (DeSE), Virtual Conference.: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/DeSE51703.2020.9450232\u003c/span\u003e\u003cspan address=\"10.1109/DeSE51703.2020.9450232\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOnah, D., Pang, E., El-Haj, M. A., \u0026amp; Data-driven (2022). Latent Semantic Analysis for Automatic Text Summarization using LDA Topic Modelling. In: Proceedings of the 2022 IEEE International Conference on Big Data (Big Data), Osaka, Japan.; : \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/bigdata55660.2022.10020259\u003c/span\u003e\u003cspan address=\"10.1109/bigdata55660.2022.10020259\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoda, N., Ahmad, N., \u0026amp; Mahmood, M. R. (2022). Students\u0026rsquo; Satisfaction with Technology-Assisted Learning: An Empirical Analysis of Female University Students in Saudi Arabia Using Telecourse Evaluation Questionnaire. In: Saini HS, Singh RK, Tariq Beg M, Mulaveesala R, Mahmood MR (Eds.) Innovations in Electronics and Communication Engineering. Lecture Notes in Networks and Systems. Springer, Singapore. ;355: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-981-16-8512-5_51\u003c/span\u003e\u003cspan address=\"10.1007/978-981-16-8512-5_51\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMay, D., Morkos, B., Jackson, A., Hunsu, N. J., Ingalls, A., \u0026amp; Beyette, F. (2022). Rapid transition of traditionally hands-on labs to online instruction in engineering courses. \u003cem\u003eEuropean Journal of Engineering Education\u003c/em\u003e, \u003cem\u003e48\u003c/em\u003e, 842\u0026ndash;860. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/03043797.2022.2046707\u003c/span\u003e\u003cspan address=\"10.1080/03043797.2022.2046707\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmashi, R., Dandgall, S. S., \u0026amp; Vijayalakshmi, M. (2023). Feedback for Faculty on Student\u0026rsquo;s Asynchronous Learning based on Classification using Topic Modelling. In: Proceedings of the IEEE Delhi Section Flagship Conference (DELCON), 2nd Edition, Rajpura, India.: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/DELCON57910.2023.10127267\u003c/span\u003e\u003cspan address=\"10.1109/DELCON57910.2023.10127267\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJones, K. (1972). A statistical interpretation of term specificity and its application in retrieval. \u003cem\u003eJournal of Documentation\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e, 11\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSalton, G., Wong, A., \u0026amp; Yang, C. (1975). A vector space model for automatic indexing. \u003cem\u003eCommunications of the ACM\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e, 613\u0026ndash;620.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLitman, D. (2016). Natural language processing for enhancing teaching and learning. In: Proceedings of the 30th AAAI conference on artificial intelligence, AAAI Press, Phoenix, Arizona, USA. :\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1609/aaai.v30i1.9879\u003c/span\u003e\u003cspan address=\"10.1609/aaai.v30i1.9879\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShaik, T., Tao, X., LiY, Dann, C., Mcdonald, J., Redmond, P., \u0026amp; Galligan, L. (2022). A Review of the Trends and Challenges in Adopting Natural Language Processing Methods for Education Feedback Analysis. \u003cem\u003eIeee Access : Practical Innovations, Open Solutions\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e, 56720\u0026ndash;56739. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2022.3177752\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2022.3177752\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGottipati, S., Shankararaman, V., \u0026amp; Lin, J. R. (2018). Text analytics approach to extract course improvement suggestions from students\u0026rsquo; feedback. \u003cem\u003eResearch and Practice in Technology Enhanced Learning\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41039-018-0073-0\u003c/span\u003e\u003cspan address=\"10.1186/s41039-018-0073-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, L., Huang, Y., Yang, X., Yu, S., \u0026amp; Zhuang, F. (2022). An automatic short-answer grading model for semi-open-ended questions. \u003cem\u003eInteractive Learning Environments\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e, 177\u0026ndash;190. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10494820.2019.1648300\u003c/span\u003e\u003cspan address=\"10.1080/10494820.2019.1648300\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo, W., \u0026amp; Litman, D. J. (2016). Determining the quality of a student reflective response. In: Proceedings of the 29th International Florida Artificial Intelligence Research Society Conference (FLAIRS), Key Largo, Florida, USA. ;226\u0026ndash;231.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEl Kassas, W. S., Salama, C. R., Rafea, A. A., \u0026amp; Mohamed, H. K. Automatic text summarization: A comprehensive survey. \u003cem\u003eExpert Systems with Applications\u003c/em\u003e.2021;165,113679:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.eswa.2020.113679\u003c/span\u003e\u003cspan address=\"10.1016/j.eswa.2020.113679\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSyed, S., \u0026amp; Spruit, M. Full-Text or Abstract? Examining Topic Coherence Scores Using Latent Dirichlet Allocation. In Proceedings of the IEEE International Conference on Data Science and Advanced Analytics (DSAA).2017;Tokyo, Japan (165\u0026ndash;174): \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/DSAA.2017.61\u003c/span\u003e\u003cspan address=\"10.1109/DSAA.2017.61\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, L., \u0026amp; VanLehn, K. (2017). Adaptively selecting biology questions generated from a semantic network. \u003cem\u003eInteractive Learning Environments\u003c/em\u003e, \u003cem\u003e25\u003c/em\u003e, 828\u0026ndash;846. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/10494820.2016.1190939\u003c/span\u003e\u003cspan address=\"10.1080/10494820.2016.1190939\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalamboukis, T. Z. (1995). Suffix stripping with modern Greek. \u003cem\u003eProgram\u003c/em\u003e, \u003cem\u003e29\u003c/em\u003e, 313\u0026ndash;321.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTambouratzis, G., \u0026amp; Carayannis, G. (2001). Automatic corpora-based stemming in Greek. \u003cem\u003eLiteracy and Linguistic Computing\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e, 445\u0026ndash;466. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/llc/16.4.445\u003c/span\u003e\u003cspan address=\"10.1093/llc/16.4.445\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNtais, G. (2006). Development of a stemmer for the Greek language. MSc Thesis, Stockholm University, Sweden.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaroukos, S. Enhancing a Greek Language Stemmer.2008;MSc Thesis, University of Tampere, Finland.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNtais, G., Saroukos, S., Berki, E., \u0026amp; Dalianis, H. (2016). Development and Enhancement of a Stemmer for the Greek Language. In: Proceedings of the 20th Pan-Hellenic Conference on Informatics (PCI '16), Association for Computing Machinery, New York, NY, USA.; : \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3003733.3003775\u003c/span\u003e\u003cspan address=\"10.1145/3003733.3003775\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePorter, M. F. (1980). An algorithm for suffix stripping. \u003cem\u003eProgram: automated library and information systems\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e, 130\u0026ndash;137.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePechlivanis, K., \u0026amp; Florou, E. (2015). : \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://bitbucket.org/dataengineering/stemming/src/master/\u003c/span\u003e\u003cspan address=\"https://bitbucket.org/dataengineering/stemming/src/master/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, Accesed 25 January 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePechlivanis, K., \u0026amp; Konstantopoulos, S. (2015). Corpus based methods for learning models of metaphor in modern Greek. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics. ; 9449:219\u0026ndash;228: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-319-25789-1_21\u003c/span\u003e\u003cspan address=\"10.1007/978-3-319-25789-1_21\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRousseeuw, P. J., \u0026amp; Silhouettes (1987). A graphical aid to the interpretation and validation of cluster analysis. \u003cem\u003eJournal of Computational and Applied Mathematics\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e, 53\u0026ndash;65. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0377-0427(87)90125-7\u003c/span\u003e\u003cspan address=\"10.1016/0377-0427(87)90125-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavies, D., \u0026amp; Bouldin, D. (1979). A cluster separation measure. \u003cem\u003eIEEE Transactions on Pattern Analysis and Machine Intelligence\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e, 224\u0026ndash;227. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/TPAMI.1979.4766909\u003c/span\u003e\u003cspan address=\"10.1109/TPAMI.1979.4766909\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCalinski, T., \u0026amp; Harabasz, J. (1974). A dendrite method for cluster analysis. \u003cem\u003eCommunications in Statistics\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e, 1\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/03610927408827101\u003c/span\u003e\u003cspan address=\"10.1080/03610927408827101\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHagen, L. (2018). Content analysis of e-petitions with topic modeling: How to train and evaluate LDA models? \u003cem\u003eInformation Processing and Management\u003c/em\u003e, \u003cem\u003e54\u003c/em\u003e, 1292\u0026ndash;1307. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.ipm.2018.05.006\u003c/span\u003e\u003cspan address=\"10.1016/j.ipm.2018.05.006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-computing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Computing](https://link.springer.com/journal/10791)","snPcode":"10791","submissionUrl":"https://submission.springernature.com/new-submission/10791/3","title":"Discover Computing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"educational text mining, topic modelling, Latent Dirichlet Allocation, e-assessment, student essays, recommendation for teachers","lastPublishedDoi":"10.21203/rs.3.rs-4387141/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4387141/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eEducational text mining is a rapidly growing field, thanks to the adoption of current probabilistic and machine learning algorithms. The current study focuses on student e-assessment through open-ended questions that require answers in the form of free text (student essays). Their analysis and evaluation are resource-demanding tasks for the instructor, even when supported by modern e-learning platforms. Topic modelling through the Latent Dirichlet Allocation algorithm is employed in an experimental setup, aiming to (a) extract meaningful topics from the body of pooled student answers (interpretable in the educational context of the course), (b) align the extracted topics to the \u0026lsquo;native\u0026rsquo; internal structure of the body of texts, and (c) produce recommendations for the teacher in the form of alternative (meaningful) restructurings of the e-assessment units and consequently of the course content units. Quantitative and qualitative evaluation of the extracted topic models yield positive results for the first two aims, while at the same time, and regarding the third aim, the extracted topic models directly recommend for the teacher possible restructurings of the course content. These recommendations are of practical use for the teacher, especially when he/she seeks to restructure a course, either by shrinking or by expansion (fewer or more internal units). In conclusion, topic modelling opens a spectrum of possibilities for the teacher interested to explore ways to improve the structure and organization of his/her course.\u003c/p\u003e","manuscriptTitle":"Text mining technologies applied to free-text answers of students in e-assessment: an experimental study in Greek","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-05-31 09:24:45","doi":"10.21203/rs.3.rs-4387141/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-07-30T08:24:28+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-07-29T11:59:57+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"252873734668970465827297753100404325986","date":"2024-07-29T07:03:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"314024172941878190491634170484963886685","date":"2024-07-10T07:53:50+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-06-23T21:10:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"21861229159898565920164210738566298973","date":"2024-06-13T21:11:44+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-06-13T03:44:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-06-04T07:22:04+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-05-20T09:04:05+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Computing","date":"2024-05-08T06:58:02+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-computing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Computing](https://link.springer.com/journal/10791)","snPcode":"10791","submissionUrl":"https://submission.springernature.com/new-submission/10791/3","title":"Discover Computing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a8c80675-09a1-4dad-ad13-730d70043c63","owner":[],"postedDate":"May 31st, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-01-20T15:59:31+00:00","versionOfRecord":{"articleIdentity":"rs-4387141","link":"https://doi.org/10.1007/s10791-024-09496-9","journal":{"identity":"discover-computing","isVorOnly":false,"title":"Discover Computing"},"publishedOn":"2025-01-17 15:57:04","publishedOnDateReadable":"January 17th, 2025"},"versionCreatedAt":"2024-05-31 09:24:45","video":"","vorDoi":"10.1007/s10791-024-09496-9","vorDoiUrl":"https://doi.org/10.1007/s10791-024-09496-9","workflowStages":[]},"version":"v1","identity":"rs-4387141","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4387141","identity":"rs-4387141","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}