The Ottoman-Turkish Transliteration using Traditional NLP Techniques

The Ottoman-Turkish Transliteration using Traditional NLP Techniques | 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 The Ottoman-Turkish Transliteration using Traditional NLP Techniques Ishak Dölek, Atakan Kurt This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5735281/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Ottoman-Turkish transliteration is a relatively new problem. To make a vast amount of historical documents, books, newspapers, and magazines accessible to a wider audience unfamiliar with the Ottoman script, it is necessary to transliterate the Ottoman script into the Latin-based Turkish script. This study employs traditional NLP techniques to develop a dictionary-based Ottoman-Turkish transliteration system. Using a dataset of 2403 sentences and 31K words, we achieved a Word Error Rate (WER) of 20.69% (raw), 6.31% (normalized) and a Character Error Rate (CER) of 6.46% (raw) 3.01% (normalized), resulting in a BLEU score of 51.90 (raw) 77.18 (normalized). The results show that the proposed system has a promising performance for Ottoman-Turkish transliteration. Ottoman-Turkish transliteration Ottoman-Turkish conversion Orthographic transliteration NLP techniques Dictionary-based transliteration Figures Figure 1 1. Introduction Ottoman is a version of Turkish heavily influenced by Arabic and Persian. It was the written language of the Ottoman Empire. It was both the official and literary language. At the time Turkish was the language of the land, Arabic of the religion, and Persian of the literature. Six centuries of foreign influence had affected lexicon, phonology, morphology, and syntax. There are many foreign words and structures in Ottoman so that ordinary people of today can’t understand it to a large extent today. The empire left behind millions of documents in the form of books, magazines, newspapers, and official records of all kinds stored away in Ottoman archives, libraries, and other types of repositories. They comprise one of the biggest cultural heritage collections in the world. Since the Ottoman alphabet was replaced with a new Latin-based Turkish alphabet in 1928, ordinary people couldn’t even read Ottoman let alone understand it. Hence, the conversion of Ottoman documents to modern Turkish remains as one of the greatest challenges in Digital Ottoman Studies (DOS) nowadays. Obviously there is a need to transfer these documents to modern Turkish. The transfer can be performed in three distinct steps of Ottoman OCR (or called transcription in digital humanities), Ottoman-Turkish transliteration, and Ottoman-Turkish translation. Osmanlica.com: End-To-End Conversion of Ottoman Documents to Modern Turkish ongoing project aims to convert Arabic-based Ottoman to Latin-based Modern Turkish [ 1 ]. Transliteration is the process of rewriting texts in a different alphabet [ 2 ] [ 3 ]. Usually, transliteration takes place between non-Latin-based alphabets such as Chinese, Japanese, Korean, Indian, Arabic, Persian, Urdu, and Latin-based alphabets such as English, French, Spanish, German, etc. The terms Latinization or Romanization are also used when the target language is a Latin-based one. There may be different reasons for transliteration of documents such as translation to a different language, preparing scholarly versions of historical or literary works, learning a new language, or a new alphabet. The purpose of Ottoman-Turkish transliteration is to enable people who know the Turkish alphabet to read Ottoman text originally written in the Ottoman alphabet. Later on, this text can be used to translate Ottoman to modern Turkish via machine translation. The paper is organized as follows: Section 2 reviews related work. Section 3 describes the proposed Ottoman-Turkish transliteration system. Experiment results are presented in Section 4 . Finally, Section 5 summarizes the findings and concludes the paper. 2. Related Work Limited research has been conducted on Ottoman-Turkish transliteration. Ottoman-Turkish transliteration research can be categorized into two main approaches: deep learning-based and rule-based machine transliteration. Before delving into the specific domain of Ottoman-Turkish transliteration, we briefly review some general studies in machine transliteration.[ 4 ] presents an algorithm for transliterating Arabic names into Latin script. [ 5 ] proposes a statistical transliteration model for English-Arabic cross-language information retrieval. [ 6 ] focus on the out-of-the-vocabulary (OOV) problem for foreign names in NLP by utilizing a phoneme-based transliteration approach, improving handling of proper nouns across language.[ 7 ] explores deep learning for transliteration, replacing traditional rule-based methods with data-driven models trained on bilingual datasets to achieve superior accuracy. [ 8 ] investigates transliteration of Romanized Assamese social media text, proposing a machine transliteration framework that accommodates informal styles and phonetic variations typical of social media usage Studies[ 9 ] [ 10 ][ 11 ] provide examples of rule-based systems. For instance, a study[ 9 ] aims to develop an automated system for transcribing Ottoman texts into modern Turkish script. Key contributions of this study include: (a) a detailed description of regular Ottoman orthography, (b) an analysis of script-specific challenges, exceptions, and variations, and (c) a proposed transliteration framework and system architecture.[ 10 ] introduces a rule-based system for transliterating Arabic-based Ottoman into Latin-based Turkish. The system employs partial morphological parsing of Ottoman words, utilizing dictionaries to identify root words. The Turkish transliteration is generated by conjugating and declining the parsed morphology into contemporary spelling and pronunciation. Two dictionaries are used in this approach: one for Arabic and Persian loanwords and another for Modern Turkish vocabulary. While the search in the loanword dictionary is direct, the system derives patterns from the Ottoman spelling to facilitate searches in the modern dictionary. Another study[ 11 ] involves multiple stages to address the complexities of transliteration. It begins with basic character mapping for regular pronunciation and orthographic correspondence, complemented by rule-based approaches and normalization to handle irregular and exceptional cases. The system demonstrates an overall accuracy of 73.9% To our knowledge, there are no existing studies that employ deep learning techniques for direct Ottoman-Turkish transliteration. However, some studies have explored direct conversion of Ottoman images to Turkish text using OCR or HTR techniques [ 12 ] [ 13 ] [ 14 ]. [ 13 ] employs deep learning techniques for automated transcription of late 19th- and early 20th-century periodicals written in Arabic-script Ottoman using the Transkribus platform. [ 14 ] addresses the challenge of automatic transcription of printed Ottoman documents. The system’s performance was measured in terms of 6.59% - CER and 28.46%- WER on a test set of 6,828 text lines. 3. Ottoman-Turkish Transliteration Method Ottoman-Turkish transliteration remains a tremendously challenging and complex task due to its unique orthography, historical context, and the lack of a standardized system. Addressing the challenges posed by Ottoman-Turkish transliteration necessitates a multifaceted approach, integrating techniques from traditional natural language processing such as morphological analysis, synthesis, word segmentation, spelling correction, word prediction (language models) and noun phrases and compound words. Through the synergistic combination of these components, Ottoman scripts can be effectively transformed into meaningful Turkish equivalents. Turkish is a phonetic language, while Ottoman is a non-phonetic language. A phonetic language is a language that is written as it is read or read as it is written. In other words, in phonetic languages, there is a one-to-one correspondence between the letters in the alphabet and the sounds in the language. To put it more clearly, a letter in writing expresses only one sound in the language, and a sound in the language is expressed by only one letter in writing. The two main reasons why Ottoman is not phonetic are (i) the fact that vowels are mostly not written and (ii) a letter can correspond to more than one sound. Since Ottoman is not phonetic “بر → bir, ber, birr; كل → gül, gel, kel, kil, gil, kül; ایلە → ile, ayla, eyle”, as in the examples of different words can have the same homographs. Therefore, when reading Ottoman texts, ambiguity arises in such words and this situation makes both reading Ottoman and Ottoman-Turkish transliteration difficult. In summary, Ottoman-Turkish transliteration presents several challenges due to the following factors: Arabic and Persian words that are not written with vowels in their original form Words that are commonly written without vowels Suffixes that are traditionally written without vowels Words where vowels are sometimes omitted. Considering these factors, several examples are provided below. These factors pose significant challenges for Ottoman-Turkish transliteration and directly impact the WER, CER, and BLEU score. Words ending in (ة) in Ottoman Turkish can be read as Turkish (en): حقیقة → hakikaten, مادةً →maddeten The spelling and pronunciation of a word in Ottoman may be different from each other: → kanbur (kambur), پنجشنبه → pencşenbe (perşembe), هینجغیرەرق→ hıncğırarak (hıçkırarak) etc. Two words in Ottoman can be combined to form different words: او قدر → o kadar when these words merged اوقدر→ oktur this word occur. Both stems and suffixes in Turkish follow the vowel harmony rule. When suffixes are added to loanwords in Ottoman, it does not comply with vowel harmony. كوفیلرە → kufilara (kufilere) The root of a word is preserved in Ottoman, but in some cases, some letters are omitted. ایلر (ایلەر) → eyler, سویلر (سویلەر)→ söyler A word can be broken up into two meaningful words in Ottoman. سنك كلمە → kelime senin (kelimesinin), سندن كلمە → gelme senden (gelmesinden), سنده مختلفە → muhtelife sende (muhtelifesinde) As shown in Table 1 , There are many-to-many, 1-to-many, and 1-to-1 mappings between Ottoman and Turkish alphabets. There are some many-to-one mappings between characters and phonemes in the Ottoman alphabet. As opposed to, the Turkish alphabet has only one-to-one mappings between characters and phonemes. The Ottoman letter ‘ش’ corresponds to the Turkish letter ‘ş’, while ‘ی’ corresponds to ‘y, i, ı’. The Ottoman letter ‘ا’ can correspond to either ‘a’ or ‘e’ in Turkish, and ‘و’ can correspond to ‘v,’ ‘o,’ ‘ö,’ ‘u,’ or ‘ü’. As mentioned before, only consonants and long vowels are generally written in Ottoman. Vowels are sometimes omitted in many word structures. Some example is shown in Table 2 . Considering such situations, Ottoman-Turkish transliteration is a challenging and complex task. This study presents a six-step system for Ottoman-Turkish transliteration, as illustrated in Fig. 1 . The steps are outlined below: Orthographic transliteration Word segmentation Spelling correction Vowelization Word prediction Noun phrases and Compound words This study focuses specifically on the orthographic transliteration step within the Ottoman - Turkish transliteration system. Other steps of the system will not be addressed in this study. Table 1 Ottoman-Turkish Transliteration Alphabet Ottoman Ortographic transliteration Phonetic transcription Char IPA ALA-LC JIMES ا‎ ء‎ ب‎ پ‎ ت‎ ث‎ ج‎ چ‎ ح‎ خ‎ د‎ ذ‎ ر‎ ز‎ ژ‎ س‎ ش‎ ص‎ ض‎ ط‎ ظ‎ ع‎ غ‎ ف‎ ق‎ ك‎ گ‎ ڭ‎ ل‎ م‎ ن‎ و‎ ه‎ ی‎ a|e —|'|’|ʾ b|p p t s c ç h h d z r z j s ş s z|d t|d z —|'|‘|ʿ g|ğ f k k g|ğ n l m n v|o|ö|u|ü h|a|e y|ı|i æ|e ʔ b|p p t s d͡ʒ t͡ʃ h x d z ɾ z ʒ s ʃ s z|d t|d z ʔ g|j f k k g|j ŋ l m n v|o|œ|u|y h|æ|e j|ɯ|i ā|' ʾ b p t s̠ c ç ḥ ḫ d z̠ r z j s ș ṣ ż ṭ ẓ ʿa ġ f q k g ñ l m n v|ū|aw|avv|ūv h y|ī|ay|á|īy a|e ʾ b p t s̠ c ç ḥ ḫ d z̠ r z j s ș ṣ ż ṭ ẓ ʿa g|ğ f ḳ k|g|ğ|y g ñ l m n v h y 34 31 31 43 34 Table 2 Ottoman-Turkish vowels representation Origin Ottoman script Turkish script Meaning Description Arabic مثلا mesela example e is omitted Arabic محكمە mahkeme court a and e are omitted Arabic ذلت zillet humiliation i and e are omitted Persian آسمان asuman sky u is omitted Persian آتش ateş fire e is omitted Persian دل dil heart i is omitted Turkish صقال sakal beard a is omitted Turkish تمل temel basis e is omitted Turkish كچوك küçük small ü is omitted Orthographic transliteration is performed in three steps: morphological parsing, transliteration dictionary, and morphological synthesis. Morphological parsing it is the process of identifying the morphemes (stems and suffixes) that constitute a given word in NLP, which involves extracting potential stem and suffix pairs from Ottoman words in this step. Transliteration dictionary To find the corresponding Turkish stem and suffix pairings, the transliteration dictionary is utilized to search for extracted stem and suffix pairs in bilingual stem and suffix dictionaries. Morphological synthesis it is the reconstruction of possible Turkish words using transliteration stem and suffix pairs. The accuracy of Ottoman-Turkish transliteration heavily relies on a comprehensive the transliteration dictionary. To this end, we have integrated over 10 widely used lexicons, including Kamus-i Turki, Kamus-i Osmani, Lehce-i Osmani, Vankulu, Lugati Naci, Lugat-i, and Lugat-i Remzi. The transliteration dictionary encompasses about 126,000 words. Examples of stem and suffix dictionaries are provided in Table 3 and Table 4. The proposed system initially processes an input Ottoman text by segmenting it into sentences and subsequently into individual words. Transliteration is then applied to each word sequentially. The word "عثمانليلاشديرامايابيله جكلريمزدنمشسڭزجه سنه" exemplifies the complexity of Ottoman words, often comprising multiple morphemes (stems and suffixes). Accurate parsing, dictionary lookup, and synthesis are crucial for successful transliteration of such complex words. Table 5 presents several examples. The provided morphological parsing result correctly identifies: Stem: عثمان Suffixes: لی+لاش+دير+امايابل+ەجك+لريمز+دن+مش+سڭز+جه+سنه The result of searching in transliterations dictionary: Stem: عثمان → Osman Suffixes: لی → lH, لاش→ lAş, دير→ DHr, امايابل→ AmAyAbAl, ەجك → AcAk, لريمز→ lArHmHz, دن → DAn, مش→ mHş, سڭز→ sHnHz, جه→ cA, سنه→ sHnA The synthesizing results: Stem: Osman Suffixes: lı+laş+tır + amayabil + ecek + lerimiz + den + miş+siniz + ce + sine Result: osmanlılaştıramayabileceklerimizdenmişsinizcesine Table 3:Ottoman and Turkish script stems Ottoman script Turkish script ابرد ebred ابرح ebrah ابرج ebrec عبرت ibret ابرايل ebrail ابراهيم ibrahim عبرخ âbrâh ابرار ebrâr ابدي ebedî ابدال ibdâl Table 4: Ottoman and Turkish script suffixes Ottoman script Turkish script Description لغی lHğH H → {i, i, u, ü} A → {a, e} D → {d, t} سل sAl مڭدر mHnDHr داش dAş چه çA Table 5 Ottoman-Turkish transliteration examples Ottoman script Transliteration Dictionary Turkish script Morphological parsing Morphological synthesis word stem + suffixes word + suffix/es word باشلا باش +لا baş+la başla گلمك گل+مك gül/gel + mek gülmek + gelmek بردر بر+در bir + dir birdir چركينلكلري چركين+لكلري çirkin + likleri çirkinlikleri الڭزده كي ال+ڭزده كي el + inizdeki elinizdeki لياقتسزلگمزي لياقت+سزلگمزي liyakat + sizliğimiz liyakatsizliğimiz عثمانليلاشديرامايابيله جكلريمزدنمشسڭزجه سنه osmanlılaştıramayabileceklerimizdenmişsinizcesine The following algorithm (Table 6 ) is used for morphological parsing, transliteration dictionary and morphological synthesis steps. Morphological parsing : An iterative approach is used to identify potential stems and suffixes within the Ottoman word. Starting from the end of the word, one letter is removed at a time, and the resulting substring is checked against the dictionary. This process continues until a valid stem or suffix is found or the word is reduced to a single letter. Transliteration Dictionary Lookup : Identified stems and suffixes are mapped with their corresponding Turkish equivalents. Morphological Synthesis : The Turkish equivalents of the identified stems and suffixes are combined to form potential Turkish words. These synthesized words are filtered based on Turkish grammatical rules to eliminate invalid combinations. The remaining valid Turkish words represent the possible transliterations of the original Ottoman word. Table 6 Ottoman-Turkish transliteration algorithm ALGORITHM1: Ottoman-Turkish transliteration transliteration (ottoman_text, ottoman_lexicon): result ← [] unique_words ← ottoman_text word_pointer ← len(unique_words) – 1 while word_pointer > 0 ottoman_word ← unique_words[word_pointer] word_index ← len(ottoman_word) − 1 ottoman_suffix ← "" while word_index > 1 if ottoman_word in ottoman_lexicon ottoman_stem, turkish_stem = ottoman_lexicon[ottoman_word] if ottoman_suffix == "" result.append(ottoman_stem, turkish_stem) else ottoman_word, turkish_word = merge_stem_and_suffix(ottoman_word, ottoman_suffix) result.append(ottoman_word, turkish_word)) word_index ← word_index − 1 ottoman_word ← ottoman_word[: word_index] ottoman_suffix ← ottoman_word[word_index:] word_pointer ← word_pointer − 1 return result 4. Dataset, Experimental and Results 4.1 Test dataset The quality of the test dataset has a significant impact on the efficacy of the system designed for Ottoman-Turkish transliteration using traditional NLP approaches. A well-constructed test dataset enables a more accurate evaluation of model performance, identification of weaknesses, and subsequent improvements, leading to a more reliable and nuanced understanding of our historical and cultural heritage. To evaluate the proposed Ottoman-Turkish transliteration system, a parallel Ottoman-Turkish test dataset was created consisting of approximately 100 pages (Table 7 ). The examples of sentence from the test dataset is provided in Table 8 . This dataset was assembled by selecting 100 distinct Ottoman documents (books) and extracting a single page from each, ensuring a diverse representation of Ottoman texts. This approach maximizes the variety of words within the test dataset, leading to a more accurate and realistic assessment of the system's performance. Additionally, the distribution of sentence length for the test dataset is illustrated in Table 9 . The test dataset is readily accessible at osmanlica.com/en/test. Table 7 Ottoman-Turkish test dataset Ottoman sentence 2.403 Ottoman words 27K Ottoman letters 192K Turkish sentence 2.403 Turkish words 27K Turkish letters 225K Table 8 Ottoman-Turkish test dataset examples # Ottoman Turkish 1 هپمز ياپاجغمز شيئي بيلييوردك Hepimiz yapacağımız şeyi biliyorduk 2 بو اولنەجگمز هفته ايدي Bu evleneceğimiz hafta idi 3 بن ده منّتدار موافقت ايتدم Ben de minnettar muvafakat ettim 4 آرقەسنده گورمەمي ايستەيوردي Arkasında görmemi istiyordu 5 چاغيريرز كيمسه امداديمزه گلمز Çağırırız kimse imdadımıza gelmez 6 آصيل اولدقلرينه داها چوق بڭزرلر Asıl olduklarına daha çok benzerler 7 كه آندن اولمەيه آزرده بر دل Ki andan olmaya azürde bir dil 8 الباس ايدر و سائره ilbas eder ve saire 9 و اونڭله اشتغال ايدر ve onunla iştigal eder 10 بو اعتبارله اركاني حربيه عموميه رياستي مستقلدر Bu itibarla Erkanı Harbiye-i Umumiye Riyaseti müstakildir Table 9 Test dataset sentence distribution by length Sentence length % Sentence length % 1 2.35 13 3.65 2 3.32 14 3.28 3 3.97 15 3.05 4 5.68 16 2.40 5 5.22 17 2.35 6 7.11 18 2.45 7 7.57 19 2.12 8 6.46 20 1.62 9 6.33 21 1.75 10 6.05 22 1.85 11 5.54 23 1.20 12 4.11 24 1.20 4.2 Experiment Results Various metrics, such as CER, WER, and BLEU score, can be employed to evaluate the accuracy of machine transliteration systems [ 17 ]. CER metrics, which measure character-level errors, are used in transliteration evaluation studies. However, in highly context-sensitive situations such as Ottoman-Turkish transliteration, WER metrics, which measure word-level errors, are a more appropriate option. The accuracy of the transliteration of words is the main objective of such studies. Since Ottoman words can have several Turkish equivalents, context analysis and subsequent word prediction play an important role in evaluating the performance of transliteration systems. In this study, WER and BLEU scores and CER values are calculated to assess the transliteration system from different aspects. Due to its long historical usage, the Ottoman exhibits significant variation in the spelling and pronunciation of words over time. These inconsistencies between graphemes and phonetics (as illustrated in Table 10 ) can significantly impact the accuracy assessment of text processing systems. To address this challenge, a normalization process was applied to standardize these variant word forms. By normalizing the words, we were able to calculate more reliable accuracy rates for the texts. Table 10 Raw and Normalized word examples Ottoman Turkish (Raw) Turkish (Normalized) 1 شمدی imdi, şimdi şimdi 2 قيلور kılur, kılır kılır 3 گلور gelür, gelir gelir 4 گيجەلري giceleri, geceleri geceleri 5 ايتديريلمه يوب itdirilmeyüb ettirilmeyip 6 قانبور kanbur kambur 7 پنجشنبه pencşenbe perşembe 8 ايدن iden eden 9 بويورمشدر buyurmuşdur buyurmuştur 10 چونكه çünki, çünkü To evaluate the WER and visualize word-level transliterations or replacements, we utilize the evaluate library [ 18 ]. WER is a widely used metric for quantifying the difference between a reference (ground truth) sentence and a hypothesis (predicted) sentence. It is calculated as follows: $$\:WER=\frac{S+D+I}{N}$$ S = Substitutions (wrong word) D = Deletions (missing word) I = Insertions (extra word) N = Number of words in the reference sentence The proposed Ottoman-Turkish transliteration system was evaluated on a test dataset, and its performance was is given in Table 11 . The WER of 20.69% for the raw data indicates a significant level of word-level errors, likely due to challenges in recognizing or translating words accurately. However, this error rate was substantially reduced to 6.31% after normalization, demonstrating the effectiveness of the normalization process in improving the system's performance. CER 6.46% reflects relatively fewer errors at the character level compared to word-level errors. 3.01% indicates improved accuracy after normalization. BLEU Score 51.90 shows moderate alignment between the predicted and reference sequences, typically in machine translation or text generation. 77.18 indicates the system achieves a significantly higher score after normalization. The normalized F1 score of 96.60% and precision of 93.43% suggest the system is highly accurate after normalization. Normalization greatly improves performance across all metrics. The WER and CER reductions highlight the model's reliance on context cleanup, and the BLEU score improvement shows better alignment with expected outputs after accounting for normalization. Table 12 shows Ottoman, Turkish and the outputs of Ottoman-Turkish transliteration system. In addition, this system has been deployed as an online service, allowing remote users to access and utilize its capabilities at https://www.osmanlica.com/en/transliteration . Table 11 Result's Ottoman-Turkish transliteration Metric Raw % Normalized % WER 20.69 6.31 CER 6.46 3.01 Bleu score 51.90 77.18 Precision 79.70 93.43 F1 score 88.70 96.60 In Table 11 , Substitution errors are the most prominent error type. While normalization reduces them significantly, further improvement is needed to minimize substitutions further. Deletion errors are minimal and well-handled after normalization. Insertion errors are insignificant overall, though they slightly increase during normalization. The majority of errors arise from incorrect replacements. Normalization significantly reduces all error types, especially deletions and substitutions. Table 12 WER error types Raw Normalized Deletion 1.5% (401) 0.08% (22) Substitution 19.19% (5133) 6.15% (1620) Insertion 0 0.08% (22) Table 13 Ottoman-Turkish transliteration predicts examples 1 Ottoman صالح آغانڭ اوغللرندن بري قنبوردر بكر چاوشڭ قيزي زهرا كوردر Turkish Salih Ağanın oğullarından biri kamburdur Bekir Çavuşun kızı Zehra kördür Output salih ağanın oğullarından biri kanburdur bekir çavuşun kızı zera gördür 2 Ottoman بن گورمدم فقط محمد علينڭ سويلهديگنه گوره مختارڭ قاريسینی آدی بيلينمين بر علّت سكز ييلدن بري أويله بر اويروب قويرمش او قدر قارمهقاريشيق بر حاله صوقمش كه باجاقلريني قوللرندن قوللريني باجاقلرندن آييرمهنڭ امكاني يوقمش Turkish Ben görmedim fakat Mehmet Alinin söylediğine göre muhtarın karısını adı bilinmeyen bir illet sekiz yıldan beri öyle bir evirip kıvırmış o kadar karmakarışık bir hale sokmuş ki bacaklarını kollarından kollarını bacaklarından ayırmanın imkanı yokmuş Output ben görmedim fakat mehmed alanın söylediğine göre muhtarın karısını edi bilinmin bir illet sekiz yıldan beri öyle bir evirip koyarmış o kadir karmekaryşyk bir hala sokmuş ki bacaklarını kollarından kollarını bacaklarından ayırmanın imkanı yokmuş 3 Ottoman بتون وجودنده جانلي يالڭز بر يري قالمش Turkish Bütün vücudunda canlı yalnız bir yeri kalmış Output bütün vücutunda canlı yalnız bir yeri kalmış 4 Ottoman باري اولدي اولاجق شونلري ده قپاييویرسهڭه Turkish Bari oldu olacak şunları da kapayıversene Output bari oldu olacak şunları da kapayıversene 5 Ottoman نه ياپسهم بو چنبري يارهمييوریم Turkish Ne yapsam bu çemberi yaramıyorum Output ne yapsam bu çemberi yaramıyorum 5 Conclusion The Ottoman-Turkish transliteration task is a crucial aspect of converting the vast amount of documents within Ottoman archives. To the best of our knowledge, there are limited studies in this field. This study presents a holistic approach to Ottoman-Turkish transliteration, utilizing traditional Natural Language Processing (NLP) techniques such as morphological parsing, dictionary lookup, morphological synthesis, word segmentation, spelling correction, vowelization, word prediction. The primary focus of this study lies in morphological parsing, dictionary lookup, and morphological synthesis. To evaluate the system's performance, a 100-page dataset was prepared and made publicly available to facilitate further research in this area. We have taken into account some cases that significantly impact accuracy. Future research should focus on developing specialized dictionaries and algorithms to address these challenges and further improve performance. Declarations Author Contribution Ishak Dölek and Atakan Kurt conceptualized and designed the study. Atakan Kurt developed the dictionary-based transliteration system, conducted the experiments, and performed data analysis. Ishak Dölek curated the dataset, implemented the evaluation framework, and prepared the statistical analysis. Both authors wrote the main manuscript text and reviewed the final version of the manuscript. All authors approved the submitted version and are accountable for the work. Data Availability The publicly available dataset from Osmanlica.com (https://osmanlica.com/test - Ottoman-Turkish transliteration test dataset) was used to evaluate the system's performance. This dataset contains parallel text files in Ottoman and Turkish, providing a benchmark for assessing the accuracy of the transliteration process. References İ, Dölek, & Kurt, A. (2022). Osmanlıcadan Türkçeye Uçtan Uca Aktarım, Journal of Smart Systems Research, 3, 1, pp. 1–10. Salaev, U., Kuriyozov, E., & Gómez-Rodríguez, C. (2022). A machine transliteration tool between Uzbek, in The International Conference on Agglutinative Language Technologies as a challenge of Natural Language Processing, Koper. Yadav, M., Kumar, I., & Kumar, A. (2023). Different Models of Transliteration - A Comprehensive Review, in 2023 International Conference on Innovative Data Communication Technologies and Application (ICIDCA), Uttarakhand. Arbabi, M., Fischthal, S. M., Cheng, V. C., & Bart, E. (1994). Algorithms for Arabic name transliteration, IBM Journal of Research and Development, 38, 2, pp. 183–194. AbdulJaleel, N., & Larkey, L. S. (2003). Statistical transliteration for english-arabic cross language information retrieval, in In Proceedings of the twelfth international conference on Information and knowledge management (CIKM '03), New York. Gao, W., Wong, K. F., & Lam, W. (2005). Phoneme-Based Transliteration of Foreign Names for OOV Problem, in Natural Language Processing – IJCNLP 2005, Jeju Island. Deselaers, T., Hasan, S., Bender, O., & Ney, H. (2009). A Deep Learning Approach to Machine Transliteration, in Proceedings of the Fourth Workshop on Statistical Machine Translation. Baruah, H., Ranbir Singh, S., & Sarmah, P. (2024). Transliteration Characteristics in Romanized Assamese Language Social Media Text and Machine Transliteration, ACM Trans. Asian Low-Resour. Lang. Inf. Process., 23, 2, p. https://doi.org/10.1145/3639565 . Bilgin, E. F. (2012). Machine transliteration of Ottoman Turkish texts to modern Turkish, İstanbul: İstanbul Fatih Ün . Fen Bilimleri Enstitüsü. Korkut, J. (2019). Morphology and lexicon-based machine translation of Ottoman Turkish to Modern Turkish . Princeton University. Jaf, A. A., & Kayhan, S. K. (2021). Machine-Based Transliterate of Ottoman to Latin‐Based Script, Scientific Programming, p. 7152935. Bilgin Tasdemir, E. F. (2023). Printed Ottoman text recognition using synthetic data and data augmentation, International Journal on Document Analysis and Recognition (IJDAR), 26, no. https://doi.org/10.1007/s10032-023-00436-9 , pp. 273–287. Kirmizialtin, S., & Wrisley, D. (2020). Automated transcription of non-Latin script periodicals: a case study in the ottoman Turkish, arXiv preprint arXiv:2011.01139. Tasdemir, E. F. B., Tandoğan, Z., Akansu, S. D., Kızılırmak, F., Sen, M. U., Akcan, A., & Yanikoglu, B. (2024). Automatic Transcription of Ottoman Documents Using Deep Learning, in In International Workshop on Document Analysis Systems, Switzerland. Hládek, D., Ján, S., & Matúš, P. (2020). Survey of Automatic Spelling Correction, Electronics, vol. 9, no. https://doi.org/10.3390/electronics9101670 . Jurafsky, D., & James, H. M. (2018). N-gram language models, Speech and language processing, 23. Lee, S., Lee, J., Moon, H., Park, C., Seo, J., Eo, S., Koo, S., & Lim, H. (2023). A Survey on Evaluation Metrics for Machine Translation, Mathematics, vol. 4 (11), no. https://doi.org/10.3390/math11041006 , p. 1006. Hugging Face, 11 9 2024. [Online]. Available: https://huggingface.co/docs/evaluate/index . [Accessed 9 11 2024]. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-5735281","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":396169881,"identity":"fc62fb8f-c91f-4ca3-9b00-8dc7fe40d63c","order_by":0,"name":"Ishak Dölek","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9UlEQVRIiWNgGAWjYBACNgYeBsYGNgYZBh7+BwYfQCLsRGrhYeDhYSicARJhJmgPkpbPPCABQlr4+M8e/DijzIaHn+fswc02v7bJ8zEzMH74mIPHYRJ5yZIbzqXxSPb2JRvn9t02bGNmYJacuQ2fFh4DyYdth3kMzjOYGef23GYEamFj5sWnhf+M8U+QFvvzDOa/LXtu2xPWwpBjJrkRZAtvj4Exw4/biYS1SOSYWc4A+kXizLEEw96G28ltzIzNeP0i33/G+GZPmY0cf0/yAYMff27bzm9vPvjhIx4tqICxDUw2EKseBP6QongUjIJRMApGCgAAcy9L9iynJOoAAAAASUVORK5CYII=","orcid":"","institution":"Mina Ar-Ge Inc.","correspondingAuthor":true,"prefix":"","firstName":"Ishak","middleName":"","lastName":"Dölek","suffix":""},{"id":396169882,"identity":"0e17e8fb-e262-42e3-9231-2dce1efabeb2","order_by":1,"name":"Atakan Kurt","email":"","orcid":"","institution":"Istanbul University Cerrahpaşa","correspondingAuthor":false,"prefix":"","firstName":"Atakan","middleName":"","lastName":"Kurt","suffix":""}],"badges":[],"createdAt":"2024-12-30 12:05:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5735281/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5735281/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":72781591,"identity":"376a8506-4c7c-46fd-8415-ce65809a6bac","added_by":"auto","created_at":"2025-01-02 06:10:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":46053,"visible":true,"origin":"","legend":"\u003cp\u003eProposed Ottoman-Turkish transliteration system\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-5735281/v1/0a97a562b584980e73adde35.png"},{"id":76034241,"identity":"c3663eba-88d2-4576-9668-032b9807422e","added_by":"auto","created_at":"2025-02-11 15:47:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":912375,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5735281/v1/807cdbf9-b564-4d70-b2dc-359e4659d986.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"The Ottoman-Turkish Transliteration using Traditional NLP Techniques","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eOttoman is a version of Turkish heavily influenced by Arabic and Persian. It was the written language of the Ottoman Empire. It was both the official and literary language. At the time Turkish was the language of the land, Arabic of the religion, and Persian of the literature. Six centuries of foreign influence had affected lexicon, phonology, morphology, and syntax. There are many foreign words and structures in Ottoman so that ordinary people of today can\u0026rsquo;t understand it to a large extent today. The empire left behind millions of documents in the form of books, magazines, newspapers, and official records of all kinds stored away in Ottoman archives, libraries, and other types of repositories. They comprise one of the biggest cultural heritage collections in the world. Since the Ottoman alphabet was replaced with a new Latin-based Turkish alphabet in 1928, ordinary people couldn\u0026rsquo;t even read Ottoman let alone understand it. Hence, the conversion of Ottoman documents to modern Turkish remains as one of the greatest challenges in Digital Ottoman Studies (DOS) nowadays. Obviously there is a need to transfer these documents to modern Turkish. The transfer can be performed in three distinct steps of Ottoman OCR (or called transcription in digital humanities), Ottoman-Turkish transliteration, and Ottoman-Turkish translation. Osmanlica.com: End-To-End Conversion of Ottoman Documents to Modern Turkish ongoing project aims to convert Arabic-based Ottoman to Latin-based Modern Turkish [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTransliteration is the process of rewriting texts in a different alphabet [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Usually, transliteration takes place between non-Latin-based alphabets such as Chinese, Japanese, Korean, Indian, Arabic, Persian, Urdu, and Latin-based alphabets such as English, French, Spanish, German, etc. The terms Latinization or Romanization are also used when the target language is a Latin-based one. There may be different reasons for transliteration of documents such as translation to a different language, preparing scholarly versions of historical or literary works, learning a new language, or a new alphabet.\u003c/p\u003e \u003cp\u003eThe purpose of Ottoman-Turkish transliteration is to enable people who know the Turkish alphabet to read Ottoman text originally written in the Ottoman alphabet. Later on, this text can be used to translate Ottoman to modern Turkish via machine translation.\u003c/p\u003e \u003cp\u003eThe paper is organized as follows: Section 2 reviews related work. Section 3 describes the proposed Ottoman-Turkish transliteration system. Experiment results are presented in Section \u003cspan refid=\"Sec4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. Finally, Section 5 summarizes the findings and concludes the paper.\u003c/p\u003e"},{"header":"2. Related Work","content":"\u003cp\u003eLimited research has been conducted on Ottoman-Turkish transliteration. Ottoman-Turkish transliteration research can be categorized into two main approaches: deep learning-based and rule-based machine transliteration. Before delving into the specific domain of Ottoman-Turkish transliteration, we briefly review some general studies in machine transliteration.[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] presents an algorithm for transliterating Arabic names into Latin script. [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e] proposes a statistical transliteration model for English-Arabic cross-language information retrieval. [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] focus on the out-of-the-vocabulary (OOV) problem for foreign names in NLP by utilizing a phoneme-based transliteration approach, improving handling of proper nouns across language.[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e] explores deep learning for transliteration, replacing traditional rule-based methods with data-driven models trained on bilingual datasets to achieve superior accuracy. [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] investigates transliteration of Romanized Assamese social media text, proposing a machine transliteration framework that accommodates informal styles and phonetic variations typical of social media usage\u003c/p\u003e \u003cp\u003eStudies[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e][\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] provide examples of rule-based systems. For instance, a study[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] aims to develop an automated system for transcribing Ottoman texts into modern Turkish script. Key contributions of this study include: (a) a detailed description of regular Ottoman orthography, (b) an analysis of script-specific challenges, exceptions, and variations, and (c) a proposed transliteration framework and system architecture.[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] introduces a rule-based system for transliterating Arabic-based Ottoman into Latin-based Turkish. The system employs partial morphological parsing of Ottoman words, utilizing dictionaries to identify root words. The Turkish transliteration is generated by conjugating and declining the parsed morphology into contemporary spelling and pronunciation. Two dictionaries are used in this approach: one for Arabic and Persian loanwords and another for Modern Turkish vocabulary. While the search in the loanword dictionary is direct, the system derives patterns from the Ottoman spelling to facilitate searches in the modern dictionary. Another study[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] involves multiple stages to address the complexities of transliteration. It begins with basic character mapping for regular pronunciation and orthographic correspondence, complemented by rule-based approaches and normalization to handle irregular and exceptional cases. The system demonstrates an overall accuracy of 73.9%\u003c/p\u003e \u003cp\u003eTo our knowledge, there are no existing studies that employ deep learning techniques for direct Ottoman-Turkish transliteration. However, some studies have explored direct conversion of Ottoman images to Turkish text using OCR or HTR techniques [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] employs deep learning techniques for automated transcription of late 19th- and early 20th-century periodicals written in Arabic-script Ottoman using the Transkribus platform. [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] addresses the challenge of automatic transcription of printed Ottoman documents. The system\u0026rsquo;s performance was measured in terms of 6.59% - CER and 28.46%- WER on a test set of 6,828 text lines.\u003c/p\u003e"},{"header":"3. Ottoman-Turkish Transliteration Method","content":"\u003cp\u003eOttoman-Turkish transliteration remains a tremendously challenging and complex task due to its unique orthography, historical context, and the lack of a standardized system. Addressing the challenges posed by Ottoman-Turkish transliteration necessitates a multifaceted approach, integrating techniques from traditional natural language processing such as morphological analysis, synthesis, word segmentation, spelling correction, word prediction (language models) and noun phrases and compound words. Through the synergistic combination of these components, Ottoman scripts can be effectively transformed into meaningful Turkish equivalents.\u003c/p\u003e\n\u003cp\u003eTurkish is a phonetic language, while Ottoman is a non-phonetic language. A phonetic language is a language that is written as it is read or read as it is written. In other words, in phonetic languages, there is a one-to-one correspondence between the letters in the alphabet and the sounds in the language. To put it more clearly, a letter in writing expresses only one sound in the language, and a sound in the language is expressed by only one letter in writing. The two main reasons why Ottoman is not phonetic are (i) the fact that vowels are mostly not written and (ii) a letter can correspond to more than one sound. Since Ottoman is not phonetic \u0026ldquo;بر \u0026rarr; bir, ber, birr; كل \u0026rarr; g\u0026uuml;l, gel, kel, kil, gil, k\u0026uuml;l; ایلە \u0026rarr; ile, ayla, eyle\u0026rdquo;, as in the examples of different words can have the same homographs. Therefore, when reading Ottoman texts, ambiguity arises in such words and this situation makes both reading Ottoman and Ottoman-Turkish transliteration difficult. In summary, Ottoman-Turkish transliteration presents several challenges due to the following factors:\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003e\n \u003cp\u003eArabic and Persian words that are not written with vowels in their original form\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eWords that are commonly written without vowels\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eSuffixes that are traditionally written without vowels\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eWords where vowels are sometimes omitted.\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eConsidering these factors, several examples are provided below. These factors pose significant challenges for Ottoman-Turkish transliteration and directly impact the WER, CER, and BLEU score.\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003e\n \u003cp\u003eWords ending in (ة) in Ottoman Turkish can be read as Turkish (en): حقیقة \u0026rarr; hakikaten, مادةً \u0026rarr;maddeten\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThe spelling and pronunciation of a word in Ottoman may be different from each other: \u0026rarr; kanbur (kambur), پنجشنبه \u0026rarr; pencşenbe (perşembe), هینجغیرەرق\u0026rarr; hıncğırarak (hı\u0026ccedil;kırarak) etc.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eTwo words in Ottoman can be combined to form different words: او قدر \u0026rarr; o kadar when these words merged اوقدر\u0026rarr; oktur this word occur.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eBoth stems and suffixes in Turkish follow the vowel harmony rule. When suffixes are added to loanwords in Ottoman, it does not comply with vowel harmony. كوفیلرە \u0026rarr; kufilara (kufilere)\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThe root of a word is preserved in Ottoman, but in some cases, some letters are omitted. ایلر (ایلەر) \u0026rarr; eyler, سویلر (سویلەر)\u0026rarr; s\u0026ouml;yler\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eA word can be broken up into two meaningful words in Ottoman. سنك كلمە \u0026rarr; kelime senin (kelimesinin), سندن كلمە \u0026rarr; gelme senden (gelmesinden), سنده مختلفە \u0026rarr; muhtelife sende (muhtelifesinde)\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e, There are many-to-many, 1-to-many, and 1-to-1 mappings between Ottoman and Turkish alphabets. There are some many-to-one mappings between characters and phonemes in the Ottoman alphabet. As opposed to, the Turkish alphabet has only one-to-one mappings between characters and phonemes. The Ottoman letter \u0026lsquo;ش\u0026rsquo; corresponds to the Turkish letter \u0026lsquo;ş\u0026rsquo;, while \u0026lsquo;ی\u0026rsquo; corresponds to \u0026lsquo;y, i, ı\u0026rsquo;. The Ottoman letter \u0026lsquo;ا\u0026rsquo; can correspond to either \u0026lsquo;a\u0026rsquo; or \u0026lsquo;e\u0026rsquo; in Turkish, and \u0026lsquo;و\u0026rsquo; can correspond to \u0026lsquo;v,\u0026rsquo; \u0026lsquo;o,\u0026rsquo; \u0026lsquo;\u0026ouml;,\u0026rsquo; \u0026lsquo;u,\u0026rsquo; or \u0026lsquo;\u0026uuml;\u0026rsquo;. As mentioned before, only consonants and long vowels are generally written in Ottoman. Vowels are sometimes omitted in many word structures. Some example is shown in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. Considering such situations, Ottoman-Turkish transliteration is a challenging and complex task.\u003c/p\u003e\n\u003cp\u003eThis study presents a six-step system for Ottoman-Turkish transliteration, as illustrated in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. The steps are outlined below:\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003e\n \u003cp\u003eOrthographic transliteration\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eWord segmentation\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eSpelling correction\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eVowelization\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eWord prediction\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eNoun phrases and Compound words\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eThis study focuses specifically on the orthographic transliteration step within the Ottoman - Turkish transliteration system. Other steps of the system will not be addressed in this study.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eOttoman-Turkish Transliteration Alphabet\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOttoman\u003c/p\u003e\n \u003c/th\u003e\n \u003cth rowspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eOrtographic transliteration\u003c/p\u003e\n \u003c/th\u003e\n \u003cth colspan=\"3\" align=\"left\"\u003e\n \u003cp\u003ePhonetic transcription\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eChar\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eIPA\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eALA-LC\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eJIMES\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eا\u0026lrm;\u003c/p\u003e\n \u003cp\u003eء\u0026lrm;\u003c/p\u003e\n \u003cp\u003eب\u0026lrm;\u003c/p\u003e\n \u003cp\u003eپ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eت\u0026lrm;\u003c/p\u003e\n \u003cp\u003eث\u0026lrm;\u003c/p\u003e\n \u003cp\u003eج\u0026lrm;\u003c/p\u003e\n \u003cp\u003eچ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eح\u0026lrm;\u003c/p\u003e\n \u003cp\u003eخ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eد\u0026lrm;\u003c/p\u003e\n \u003cp\u003eذ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eر\u0026lrm;\u003c/p\u003e\n \u003cp\u003eز\u0026lrm;\u003c/p\u003e\n \u003cp\u003eژ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eس\u0026lrm;\u003c/p\u003e\n \u003cp\u003eش\u0026lrm;\u003c/p\u003e\n \u003cp\u003eص\u0026lrm;\u003c/p\u003e\n \u003cp\u003eض\u0026lrm;\u003c/p\u003e\n \u003cp\u003eط\u0026lrm;\u003c/p\u003e\n \u003cp\u003eظ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eع\u0026lrm;\u003c/p\u003e\n \u003cp\u003eغ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eف\u0026lrm;\u003c/p\u003e\n \u003cp\u003eق\u0026lrm;\u003c/p\u003e\n \u003cp\u003eك\u0026lrm;\u003c/p\u003e\n \u003cp\u003eگ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eڭ\u0026lrm;\u003c/p\u003e\n \u003cp\u003eل\u0026lrm;\u003c/p\u003e\n \u003cp\u003eم\u0026lrm;\u003c/p\u003e\n \u003cp\u003eن\u0026lrm;\u003c/p\u003e\n \u003cp\u003eو\u0026lrm;\u003c/p\u003e\n \u003cp\u003eه\u0026lrm;\u003c/p\u003e\n \u003cp\u003eی\u0026lrm;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ea|e\u003c/p\u003e\n \u003cp\u003e\u0026mdash;|\u0026apos;|\u0026rsquo;|ʾ\u003c/p\u003e\n \u003cp\u003eb|p\u003c/p\u003e\n \u003cp\u003ep\u003c/p\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003ec\u003c/p\u003e\n \u003cp\u003e\u0026ccedil;\u003c/p\u003e\n \u003cp\u003eh\u003c/p\u003e\n \u003cp\u003eh\u003c/p\u003e\n \u003cp\u003ed\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003er\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003ej\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003eş\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003ez|d\u003c/p\u003e\n \u003cp\u003et|d\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003e\u0026mdash;|\u0026apos;|\u0026lsquo;|ʿ\u003c/p\u003e\n \u003cp\u003eg|ğ\u003c/p\u003e\n \u003cp\u003ef\u003c/p\u003e\n \u003cp\u003ek\u003c/p\u003e\n \u003cp\u003ek\u003c/p\u003e\n \u003cp\u003eg|ğ\u003c/p\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003cp\u003el\u003c/p\u003e\n \u003cp\u003em\u003c/p\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003cp\u003ev|o|\u0026ouml;|u|\u0026uuml;\u003c/p\u003e\n \u003cp\u003eh|a|e\u003c/p\u003e\n \u003cp\u003ey|ı|i\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026aelig;|e\u003c/p\u003e\n \u003cp\u003eʔ\u003c/p\u003e\n \u003cp\u003eb|p\u003c/p\u003e\n \u003cp\u003ep\u003c/p\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003ed͡ʒ\u003c/p\u003e\n \u003cp\u003et͡ʃ\u003c/p\u003e\n \u003cp\u003eh\u003c/p\u003e\n \u003cp\u003ex\u003c/p\u003e\n \u003cp\u003ed\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003eɾ\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003eʒ\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003eʃ\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003ez|d\u003c/p\u003e\n \u003cp\u003et|d\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003eʔ\u003c/p\u003e\n \u003cp\u003eg|j\u003c/p\u003e\n \u003cp\u003ef\u003c/p\u003e\n \u003cp\u003ek\u003c/p\u003e\n \u003cp\u003ek\u003c/p\u003e\n \u003cp\u003eg|j\u003c/p\u003e\n \u003cp\u003eŋ\u003c/p\u003e\n \u003cp\u003el\u003c/p\u003e\n \u003cp\u003em\u003c/p\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003cp\u003ev|o|\u0026oelig;|u|y\u003c/p\u003e\n \u003cp\u003eh|\u0026aelig;|e\u003c/p\u003e\n \u003cp\u003ej|ɯ|i\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eā|\u0026apos;\u003c/p\u003e\n \u003cp\u003eʾ\u003c/p\u003e\n \u003cp\u003eb\u003c/p\u003e\n \u003cp\u003ep\u003c/p\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003cp\u003es̠\u003c/p\u003e\n \u003cp\u003ec\u003c/p\u003e\n \u003cp\u003e\u0026ccedil;\u003c/p\u003e\n \u003cp\u003eḥ\u003c/p\u003e\n \u003cp\u003eḫ\u003c/p\u003e\n \u003cp\u003ed\u003c/p\u003e\n \u003cp\u003ez̠\u003c/p\u003e\n \u003cp\u003er\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003ej\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003eș\u003c/p\u003e\n \u003cp\u003eṣ\u003c/p\u003e\n \u003cp\u003eż\u003c/p\u003e\n \u003cp\u003eṭ\u003c/p\u003e\n \u003cp\u003eẓ\u003c/p\u003e\n \u003cp\u003eʿa\u003c/p\u003e\n \u003cp\u003eġ\u003c/p\u003e\n \u003cp\u003ef\u003c/p\u003e\n \u003cp\u003eq\u003c/p\u003e\n \u003cp\u003ek\u003c/p\u003e\n \u003cp\u003eg\u003c/p\u003e\n \u003cp\u003e\u0026ntilde;\u003c/p\u003e\n \u003cp\u003el\u003c/p\u003e\n \u003cp\u003em\u003c/p\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003cp\u003ev|ū|aw|avv|ūv\u003c/p\u003e\n \u003cp\u003eh\u003c/p\u003e\n \u003cp\u003ey|ī|ay|\u0026aacute;|īy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ea|e\u003c/p\u003e\n \u003cp\u003eʾ\u003c/p\u003e\n \u003cp\u003eb\u003c/p\u003e\n \u003cp\u003ep\u003c/p\u003e\n \u003cp\u003et\u003c/p\u003e\n \u003cp\u003es̠\u003c/p\u003e\n \u003cp\u003ec\u003c/p\u003e\n \u003cp\u003e\u0026ccedil;\u003c/p\u003e\n \u003cp\u003eḥ\u003c/p\u003e\n \u003cp\u003eḫ\u003c/p\u003e\n \u003cp\u003ed\u003c/p\u003e\n \u003cp\u003ez̠\u003c/p\u003e\n \u003cp\u003er\u003c/p\u003e\n \u003cp\u003ez\u003c/p\u003e\n \u003cp\u003ej\u003c/p\u003e\n \u003cp\u003es\u003c/p\u003e\n \u003cp\u003eș\u003c/p\u003e\n \u003cp\u003eṣ\u003c/p\u003e\n \u003cp\u003eż\u003c/p\u003e\n \u003cp\u003eṭ\u003c/p\u003e\n \u003cp\u003eẓ\u003c/p\u003e\n \u003cp\u003eʿa\u003c/p\u003e\n \u003cp\u003eg|ğ\u003c/p\u003e\n \u003cp\u003ef\u003c/p\u003e\n \u003cp\u003eḳ\u003c/p\u003e\n \u003cp\u003ek|g|ğ|y\u003c/p\u003e\n \u003cp\u003eg\u003c/p\u003e\n \u003cp\u003e\u0026ntilde;\u003c/p\u003e\n \u003cp\u003el\u003c/p\u003e\n \u003cp\u003em\u003c/p\u003e\n \u003cp\u003en\u003c/p\u003e\n \u003cp\u003ev\u003c/p\u003e\n \u003cp\u003eh\u003c/p\u003e\n \u003cp\u003ey\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e34\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eOttoman-Turkish vowels representation\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOrigin\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOttoman script\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTurkish script\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eMeaning\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDescription\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eArabic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eمثلا\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003emesela\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eexample\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ee is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eArabic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eمحكمە\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003emahkeme\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ecourt\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ea and e are omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eArabic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eذلت\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ezillet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ehumiliation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ei and e are omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePersian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eآسمان\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003easuman\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003esky\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eu is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePersian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eآتش\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eateş\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003efire\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ee is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePersian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eدل\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003edil\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eheart\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ei is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTurkish\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eصقال\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003esakal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ebeard\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ea is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTurkish\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eتمل\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003etemel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ebasis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ee is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTurkish\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eكچوك\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ek\u0026uuml;\u0026ccedil;\u0026uuml;k\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003esmall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026uuml; is omitted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eOrthographic transliteration is performed in three steps: morphological parsing, transliteration dictionary, and morphological synthesis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMorphological parsing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eit is the process of identifying the morphemes (stems and suffixes) that constitute a given word in NLP, which involves extracting potential stem and suffix pairs from Ottoman words in this step.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTransliteration dictionary\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo find the corresponding Turkish stem and suffix pairings, the transliteration dictionary is utilized to search for extracted stem and suffix pairs in bilingual stem and suffix dictionaries.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMorphological synthesis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eit is the reconstruction of possible Turkish words using transliteration stem and suffix pairs.\u003c/p\u003e\n\u003cp\u003eThe accuracy of Ottoman-Turkish transliteration heavily relies on a comprehensive the transliteration dictionary. To this end, we have integrated over 10 widely used lexicons, including Kamus-i Turki, Kamus-i Osmani, Lehce-i Osmani, Vankulu, Lugati Naci, Lugat-i, and Lugat-i Remzi. The transliteration dictionary encompasses about 126,000 words. Examples of stem and suffix dictionaries are provided in Table\u0026nbsp;3 and Table\u0026nbsp;4.\u003c/p\u003e\n\u003cp\u003eThe proposed system initially processes an input Ottoman text by segmenting it into sentences and subsequently into individual words. Transliteration is then applied to each word sequentially. The word \u0026quot;عثمانليلاشديرامايابيله جكلريمزدنمشسڭزجه سنه\u0026quot; exemplifies the complexity of Ottoman words, often comprising multiple morphemes (stems and suffixes). Accurate parsing, dictionary lookup, and synthesis are crucial for successful transliteration of such complex words. Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e presents several examples.\u003c/p\u003e\n\u003cp\u003eThe provided morphological parsing result correctly identifies:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eStem: عثمان\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eSuffixes: لی+لاش+دير+امايابل+ەجك+لريمز+دن+مش+سڭز+جه+سنه\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe result of searching in transliterations dictionary:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eStem: عثمان \u0026rarr; Osman\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eSuffixes: لی \u0026rarr; lH, لاش\u0026rarr; lAş, دير\u0026rarr; DHr, امايابل\u0026rarr; AmAyAbAl, ەجك \u0026rarr; AcAk, لريمز\u0026rarr; lArHmHz, دن \u0026rarr; DAn, مش\u0026rarr; mHş, سڭز\u0026rarr; sHnHz, جه\u0026rarr; cA, سنه\u0026rarr; sHnA\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe synthesizing results:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eStem: Osman\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eSuffixes: lı+laş+tır\u0026thinsp;+\u0026thinsp;amayabil\u0026thinsp;+\u0026thinsp;ecek\u0026thinsp;+\u0026thinsp;lerimiz\u0026thinsp;+\u0026thinsp;den\u0026thinsp;+\u0026thinsp;miş+siniz\u0026thinsp;+\u0026thinsp;ce\u0026thinsp;+\u0026thinsp;sine\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eResult: osmanlılaştıramayabileceklerimizdenmişsinizcesine\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003cp\u003eTable 3:Ottoman and Turkish script stems\u003c/p\u003e\n \u003ctable border=\"1\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eOttoman script\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eTurkish script\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابرد\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eebred\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابرح\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eebrah\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابرج\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eebrec\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eعبرت\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eibret\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابرايل\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eebrail\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابراهيم\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eibrahim\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eعبرخ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003e\u0026acirc;br\u0026acirc;h\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابرار\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eebr\u0026acirc;r\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابدي\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eebed\u0026icirc;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 95px;\"\u003e\n \u003cp\u003eابدال\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 107px;\"\u003e\n \u003cp\u003eibd\u0026acirc;l\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\n \u003cp\u003eTable 4: Ottoman and Turkish script suffixes\u003c/p\u003e\n \u003ctable border=\"1\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eOttoman script\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 90px;\"\u003e\n \u003cp\u003eTurkish script\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eDescription\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eلغی\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 90px;\"\u003e\n \u003cp\u003elHğH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"5\" style=\"width: 96px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eH \u0026rarr; {i, i, u, \u0026uuml;}\u003c/p\u003e\n \u003cp\u003eA \u0026rarr; {a, e}\u003c/p\u003e\n \u003cp\u003eD \u0026rarr; {d, t}\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eسل\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 90px;\"\u003e\n \u003cp\u003esAl\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eمڭدر\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 90px;\"\u003e\n \u003cp\u003emHnDHr\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eداش\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 90px;\"\u003e\n \u003cp\u003edAş\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eچه\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 90px;\"\u003e\n \u003cp\u003e\u0026ccedil;A\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\u0026nbsp;\n \u003c/div\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eOttoman-Turkish transliteration examples\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eOttoman script\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"10\" align=\"left\"\u003e\n \u003cp\u003eTransliteration Dictionary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eTurkish script\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eMorphological parsing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eMorphological \u003cem\u003esynthesis\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eword\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003estem\u0026thinsp;+\u0026thinsp;suffixes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eword\u0026thinsp;+\u0026thinsp;suffix/es\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eword\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eباشلا\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eباش +لا\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ebaş+la\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ebaşla\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eگلمك\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eگل+مك\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eg\u0026uuml;l/gel\u0026thinsp;+\u0026thinsp;mek\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eg\u0026uuml;lmek\u0026thinsp;+\u0026thinsp;gelmek\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eبردر\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eبر+در\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ebir\u0026thinsp;+\u0026thinsp;dir\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ebirdir\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eچركينلكلري\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eچركين+لكلري\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ccedil;irkin\u0026thinsp;+\u0026thinsp;likleri\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ccedil;irkinlikleri\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eالڭزده كي\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eال+ڭزده كي\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eel\u0026thinsp;+\u0026thinsp;inizdeki\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eelinizdeki\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eلياقتسزلگمزي\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eلياقت+سزلگمزي\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eliyakat\u0026thinsp;+\u0026thinsp;sizliğimiz\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eliyakatsizliğimiz\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eعثمانليلاشديرامايابيله جكلريمزدنمشسڭزجه سنه\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eosmanlılaştıramayabileceklerimizdenmişsinizcesine\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eThe following algorithm (Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e) is used for morphological parsing, transliteration dictionary and morphological synthesis steps.\u003c/p\u003e\n \u003cp\u003e\u003cem\u003eMorphological parsing\u003c/em\u003e:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eAn iterative approach is used to identify potential stems and suffixes within the Ottoman word.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eStarting from the end of the word, one letter is removed at a time, and the resulting substring is checked against the dictionary.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThis process continues until a valid stem or suffix is found or the word is reduced to a single letter.\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003e\u003cem\u003eTransliteration Dictionary Lookup\u003c/em\u003e:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eIdentified stems and suffixes are mapped with their corresponding Turkish equivalents.\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003e\u003cem\u003eMorphological Synthesis\u003c/em\u003e:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eThe Turkish equivalents of the identified stems and suffixes are combined to form potential Turkish words.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThese synthesized words are filtered based on Turkish grammatical rules to eliminate invalid combinations.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThe remaining valid Turkish words represent the possible transliterations of the original Ottoman word.\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tab4\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eOttoman-Turkish transliteration algorithm\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth colspan=\"6\" align=\"left\"\u003e\n \u003cp\u003eALGORITHM1: Ottoman-Turkish transliteration\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\" align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003etransliteration\u003c/strong\u003e(ottoman_text, ottoman_lexicon):\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"5\" align=\"left\"\u003e\n \u003cp\u003eresult \u0026larr; []\u003c/p\u003e\n \u003cp\u003eunique_words \u0026larr; ottoman_text\u003c/p\u003e\n \u003cp\u003eword_pointer \u0026larr; len(unique_words) \u0026ndash; 1\u003c/p\u003e\n \u003cp\u003ewhile word_pointer\u0026thinsp;\u0026gt;\u0026thinsp;0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"4\" align=\"left\"\u003e\n \u003cp\u003eottoman_word \u0026larr; unique_words[word_pointer]\u003c/p\u003e\n \u003cp\u003eword_index \u0026larr; len(ottoman_word) \u0026minus;\u0026thinsp;1\u003c/p\u003e\n \u003cp\u003eottoman_suffix \u0026larr; \u0026quot;\u0026quot;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"4\" align=\"left\"\u003e\n \u003cp\u003ewhile word_index\u0026thinsp;\u0026gt;\u0026thinsp;1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"3\" align=\"left\"\u003e\n \u003cp\u003eif ottoman_word in ottoman_lexicon\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eottoman_stem, turkish_stem\u0026thinsp;=\u0026thinsp;ottoman_lexicon[ottoman_word]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eif ottoman_suffix == \u0026quot;\u0026quot;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eresult.append(ottoman_stem, turkish_stem)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\n \u003cp\u003eelse\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eottoman_word, turkish_word\u0026thinsp;=\u0026thinsp;merge_stem_and_suffix(ottoman_word, ottoman_suffix)\u003c/p\u003e\n \u003cp\u003eresult.append(ottoman_word, turkish_word))\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"3\" align=\"left\"\u003e\n \u003cp\u003eword_index \u0026larr; word_index \u0026minus;\u0026thinsp;1\u003c/p\u003e\n \u003cp\u003eottoman_word \u0026larr; ottoman_word[: word_index]\u003c/p\u003e\n \u003cp\u003eottoman_suffix \u0026larr; ottoman_word[word_index:]\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"4\" align=\"left\"\u003e\n \u003cp\u003eword_pointer \u0026larr; word_pointer \u0026minus;\u0026thinsp;1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd colspan=\"5\" align=\"left\"\u003e\n \u003cp\u003ereturn result\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"4. Dataset, Experimental and Results","content":"\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Test dataset\u003c/h2\u003e \u003cp\u003eThe quality of the test dataset has a significant impact on the efficacy of the system designed for Ottoman-Turkish transliteration using traditional NLP approaches. A well-constructed test dataset enables a more accurate evaluation of model performance, identification of weaknesses, and subsequent improvements, leading to a more reliable and nuanced understanding of our historical and cultural heritage.\u003c/p\u003e \u003cp\u003eTo evaluate the proposed Ottoman-Turkish transliteration system, a parallel Ottoman-Turkish test dataset was created consisting of approximately 100 pages (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e7\u003c/span\u003e). The examples of sentence from the test dataset is provided in Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e8\u003c/span\u003e. This dataset was assembled by selecting 100 distinct Ottoman documents (books) and extracting a single page from each, ensuring a diverse representation of Ottoman texts. This approach maximizes the variety of words within the test dataset, leading to a more accurate and realistic assessment of the system's performance. Additionally, the distribution of sentence length for the test dataset is illustrated in Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e9\u003c/span\u003e. The test dataset is readily accessible at osmanlica.com/en/test.\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 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOttoman-Turkish test dataset\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOttoman sentence\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.403\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOttoman words\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27K\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOttoman letters\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e192K\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTurkish sentence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.403\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTurkish words\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27K\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTurkish letters\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e225K\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 \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOttoman-Turkish test dataset examples\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e#\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTurkish\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eهپمز ياپاجغمز شيئي بيلييوردك\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHepimiz yapacağımız şeyi biliyorduk\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\u003eبو اولنەجگمز هفته ايدي\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBu evleneceğimiz hafta idi\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\u003eبن ده منّتدار موافقت ايتدم\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBen de minnettar muvafakat ettim\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\u003eآرقەسنده گورمەمي ايستەيوردي\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eArkasında g\u0026ouml;rmemi istiyordu\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\u003eچاغيريرز كيمسه امداديمزه گلمز\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026Ccedil;ağırırız kimse imdadımıza gelmez\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\u003eآصيل اولدقلرينه داها چوق بڭزرلر\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAsıl olduklarına daha \u0026ccedil;ok benzerler\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eكه آندن اولمەيه آزرده بر دل\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eKi andan olmaya az\u0026uuml;rde bir dil\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eالباس ايدر و سائره\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eilbas eder ve saire\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eو اونڭله اشتغال ايدر\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eve onunla iştigal eder\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eبو اعتبارله اركاني حربيه عموميه رياستي مستقلدر\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBu itibarla Erkanı Harbiye-i Umumiye Riyaseti m\u0026uuml;stakildir\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 \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 9\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTest dataset sentence distribution by length\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSentence length\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e%\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSentence length\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e%\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.65\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\u003e3.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.28\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\u003e3.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.05\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\u003e5.68\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 \u003cp\u003e2.40\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\u003e5.22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.35\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\u003e7.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.45\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.62\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.75\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.85\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.20\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=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Experiment Results\u003c/h2\u003e \u003cp\u003eVarious metrics, such as CER, WER, and BLEU score, can be employed to evaluate the accuracy of machine transliteration systems [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. CER metrics, which measure character-level errors, are used in transliteration evaluation studies. However, in highly context-sensitive situations such as Ottoman-Turkish transliteration, WER metrics, which measure word-level errors, are a more appropriate option. The accuracy of the transliteration of words is the main objective of such studies. Since Ottoman words can have several Turkish equivalents, context analysis and subsequent word prediction play an important role in evaluating the performance of transliteration systems. In this study, WER and BLEU scores and CER values are calculated to assess the transliteration system from different aspects.\u003c/p\u003e \u003cp\u003eDue to its long historical usage, the Ottoman exhibits significant variation in the spelling and pronunciation of words over time. These inconsistencies between graphemes and phonetics (as illustrated in Table\u0026nbsp;\u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e10\u003c/span\u003e) can significantly impact the accuracy assessment of text processing systems. To address this challenge, a normalization process was applied to standardize these variant word forms. By normalizing the words, we were able to calculate more reliable accuracy rates for the texts.\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 10\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRaw and Normalized word examples\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTurkish (Raw)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTurkish (Normalized)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eشمدی\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eimdi, şimdi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eşimdi\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\u003eقيلور\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ekılur, kılır\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ekılır\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\u003eگلور\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003egel\u0026uuml;r, gelir\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003egelir\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\u003eگيجەلري\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003egiceleri,\u0026nbsp;geceleri\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003egeceleri\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\u003eايتديريلمه يوب\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eitdirilmey\u0026uuml;b\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eettirilmeyip\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\u003eقانبور\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ekanbur\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ekambur\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eپنجشنبه\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003epencşenbe\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eperşembe\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eايدن\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eiden\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eeden\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eبويورمشدر\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ebuyurmuşdur\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ebuyurmuştur\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eچونكه\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ccedil;\u0026uuml;nki,\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026ccedil;\u0026uuml;nk\u0026uuml;\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\u003eTo evaluate the WER and visualize word-level transliterations or replacements, we utilize the evaluate library [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. WER is a widely used metric for quantifying the difference between a reference (ground truth) sentence and a hypothesis (predicted) sentence. It is calculated as follows:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:WER=\\frac{S+D+I}{N}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eS\u0026thinsp;=\u0026thinsp;Substitutions (wrong word)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eD\u0026thinsp;=\u0026thinsp;Deletions (missing word)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eI\u0026thinsp;=\u0026thinsp;Insertions (extra word)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eN\u0026thinsp;=\u0026thinsp;Number of words in the reference sentence\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe proposed Ottoman-Turkish transliteration system was evaluated on a test dataset, and its performance was is given in Table\u0026nbsp;\u003cspan refid=\"Tab9\" class=\"InternalRef\"\u003e11\u003c/span\u003e. The WER of 20.69% for the raw data indicates a significant level of word-level errors, likely due to challenges in recognizing or translating words accurately. However, this error rate was substantially reduced to 6.31% after normalization, demonstrating the effectiveness of the normalization process in improving the system's performance. CER 6.46% reflects relatively fewer errors at the character level compared to word-level errors. 3.01% indicates improved accuracy after normalization. BLEU Score 51.90 shows moderate alignment between the predicted and reference sequences, typically in machine translation or text generation. 77.18 indicates the system achieves a significantly higher score after normalization. The normalized F1 score of 96.60% and precision of 93.43% suggest the system is highly accurate after normalization.\u003c/p\u003e \u003cp\u003eNormalization greatly improves performance across all metrics. The WER and CER reductions highlight the model's reliance on context cleanup, and the BLEU score improvement shows better alignment with expected outputs after accounting for normalization. Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e12\u003c/span\u003e shows Ottoman, Turkish and the outputs of Ottoman-Turkish transliteration system. In addition, this system has been deployed as an online service, allowing remote users to access and utilize its capabilities at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.osmanlica.com/en/transliteration\u003c/span\u003e\u003cspan address=\"https://www.osmanlica.com/en/transliteration\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\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 11\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eResult's Ottoman-Turkish transliteration\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRaw %\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNormalized %\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWER\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20.69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.31\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCER\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBleu score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e51.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e77.18\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e79.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e93.43\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eF1 score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e88.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.60\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\u003eIn Table\u0026nbsp;\u003cspan refid=\"Tab9\" class=\"InternalRef\"\u003e11\u003c/span\u003e, Substitution errors are the most prominent error type. While normalization reduces them significantly, further improvement is needed to minimize substitutions further. Deletion errors are minimal and well-handled after normalization. Insertion errors are insignificant overall, though they slightly increase during normalization. The majority of errors arise from incorrect replacements. Normalization significantly reduces all error types, especially deletions and substitutions.\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 12\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eWER error types\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRaw\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNormalized\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeletion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.5% (401)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.08% (22)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSubstitution\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19.19% (5133)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.15% (1620)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInsertion\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\u003e0.08% (22)\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 \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab11\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 13\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOttoman-Turkish transliteration predicts examples\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eصالح آغانڭ اوغللرندن بري قنبوردر بكر چاوشڭ قيزي زهرا كوردر\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTurkish\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSalih Ağanın oğullarından biri kamburdur Bekir \u0026Ccedil;avuşun kızı Zehra k\u0026ouml;rd\u0026uuml;r\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutput\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003esalih ağanın oğullarından biri kanburdur bekir \u0026ccedil;avuşun kızı zera g\u0026ouml;rd\u0026uuml;r\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eبن گورمدم فقط محمد علينڭ سويلهديگنه گوره مختارڭ قاريسینی آدی بيلينمين بر علّت سكز ييلدن بري أويله بر اويروب قويرمش او قدر قارمهقاريشيق بر حاله صوقمش كه باجاقلريني قوللرندن قوللريني باجاقلرندن آييرمهنڭ امكاني يوقمش\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTurkish\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBen g\u0026ouml;rmedim fakat Mehmet Alinin s\u0026ouml;ylediğine g\u0026ouml;re muhtarın karısını adı bilinmeyen bir illet sekiz yıldan beri \u0026ouml;yle bir evirip kıvırmış o kadar karmakarışık bir hale sokmuş ki bacaklarını kollarından kollarını bacaklarından ayırmanın imkanı yokmuş\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutput\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eben g\u0026ouml;rmedim fakat mehmed alanın s\u0026ouml;ylediğine g\u0026ouml;re muhtarın karısını edi bilinmin bir illet sekiz yıldan beri \u0026ouml;yle bir evirip koyarmış o kadir karmekaryşyk bir hala sokmuş ki bacaklarını kollarından kollarını bacaklarından ayırmanın imkanı yokmuş\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eبتون وجودنده جانلي يالڭز بر يري قالمش\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTurkish\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eB\u0026uuml;t\u0026uuml;n v\u0026uuml;cudunda canlı yalnız bir yeri kalmış\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutput\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eb\u0026uuml;t\u0026uuml;n v\u0026uuml;cutunda canlı yalnız bir yeri kalmış\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eباري اولدي اولاجق شونلري ده قپاييویرسهڭه\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTurkish\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBari oldu olacak şunları da kapayıversene\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutput\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ebari oldu olacak şunları da kapayıversene\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOttoman\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eنه ياپسهم بو چنبري يارهمييوریم\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTurkish\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNe yapsam bu \u0026ccedil;emberi yaramıyorum\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutput\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ene yapsam bu \u0026ccedil;emberi yaramıyorum\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"},{"header":"5 Conclusion","content":"\u003cp\u003eThe Ottoman-Turkish transliteration task is a crucial aspect of converting the vast amount of documents within Ottoman archives. To the best of our knowledge, there are limited studies in this field. This study presents a holistic approach to Ottoman-Turkish transliteration, utilizing traditional Natural Language Processing (NLP) techniques such as morphological parsing, dictionary lookup, morphological synthesis, word segmentation, spelling correction, vowelization, word prediction. The primary focus of this study lies in morphological parsing, dictionary lookup, and morphological synthesis. To evaluate the system's performance, a 100-page dataset was prepared and made publicly available to facilitate further research in this area. We have taken into account some cases that significantly impact accuracy. Future research should focus on developing specialized dictionaries and algorithms to address these challenges and further improve performance.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eIshak D\u0026ouml;lek and Atakan Kurt conceptualized and designed the study. Atakan Kurt developed the dictionary-based transliteration system, conducted the experiments, and performed data analysis. Ishak D\u0026ouml;lek curated the dataset, implemented the evaluation framework, and prepared the statistical analysis. Both authors wrote the main manuscript text and reviewed the final version of the manuscript. All authors approved the submitted version and are accountable for the work.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe publicly available dataset from Osmanlica.com (https://osmanlica.com/test - Ottoman-Turkish transliteration test dataset) was used to evaluate the system's performance. This dataset contains parallel text files in Ottoman and Turkish, providing a benchmark for assessing the accuracy of the transliteration process.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eİ, D\u0026ouml;lek, \u0026amp; Kurt, A. (2022). Osmanlıcadan T\u0026uuml;rk\u0026ccedil;eye U\u0026ccedil;tan Uca Aktarım, Journal of Smart Systems Research, 3, 1, pp. 1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSalaev, U., Kuriyozov, E., \u0026amp; G\u0026oacute;mez-Rodr\u0026iacute;guez, C. (2022). A machine transliteration tool between Uzbek, in The International Conference on Agglutinative Language Technologies as a challenge of Natural Language Processing, Koper.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYadav, M., Kumar, I., \u0026amp; Kumar, A. (2023). Different Models of Transliteration - A Comprehensive Review, in 2023 International Conference on Innovative Data Communication Technologies and Application (ICIDCA), Uttarakhand.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArbabi, M., Fischthal, S. M., Cheng, V. C., \u0026amp; Bart, E. (1994). Algorithms for Arabic name transliteration, IBM Journal of Research and Development, 38, 2, pp. 183\u0026ndash;194.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdulJaleel, N., \u0026amp; Larkey, L. S. (2003). Statistical transliteration for english-arabic cross language information retrieval, in In Proceedings of the twelfth international conference on Information and knowledge management (CIKM '03), New York.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGao, W., Wong, K. F., \u0026amp; Lam, W. (2005). Phoneme-Based Transliteration of Foreign Names for OOV Problem, in Natural Language Processing \u0026ndash; IJCNLP 2005, Jeju Island.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeselaers, T., Hasan, S., Bender, O., \u0026amp; Ney, H. (2009). A Deep Learning Approach to Machine Transliteration, in Proceedings of the Fourth Workshop on Statistical Machine Translation.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaruah, H., Ranbir Singh, S., \u0026amp; Sarmah, P. (2024). Transliteration Characteristics in Romanized Assamese Language Social Media Text and Machine Transliteration, ACM Trans. Asian Low-Resour. Lang. Inf. Process., 23, 2, p. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3639565\u003c/span\u003e\u003cspan address=\"10.1145/3639565\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBilgin, E. F. (2012). \u003cem\u003eMachine transliteration of Ottoman Turkish texts to modern Turkish, İstanbul: İstanbul Fatih \u0026Uuml;n\u003c/em\u003e. Fen Bilimleri Enstit\u0026uuml;s\u0026uuml;.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKorkut, J. (2019). \u003cem\u003eMorphology and lexicon-based machine translation of Ottoman Turkish to Modern Turkish\u003c/em\u003e. Princeton University.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJaf, A. A., \u0026amp; Kayhan, S. K. (2021). Machine-Based Transliterate of Ottoman to Latin‐Based Script, Scientific Programming, p. 7152935.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBilgin Tasdemir, E. F. (2023). Printed Ottoman text recognition using synthetic data and data augmentation, International Journal on Document Analysis and Recognition (IJDAR), 26, no. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10032-023-00436-9\u003c/span\u003e\u003cspan address=\"10.1007/s10032-023-00436-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, pp. 273\u0026ndash;287.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKirmizialtin, S., \u0026amp; Wrisley, D. (2020). Automated transcription of non-Latin script periodicals: a case study in the ottoman Turkish, arXiv preprint arXiv:2011.01139.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTasdemir, E. F. B., Tandoğan, Z., Akansu, S. D., Kızılırmak, F., Sen, M. U., Akcan, A., \u0026amp; Yanikoglu, B. (2024). Automatic Transcription of Ottoman Documents Using Deep Learning, in In International Workshop on Document Analysis Systems, Switzerland.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHl\u0026aacute;dek, D., J\u0026aacute;n, S., \u0026amp; Mat\u0026uacute;š, P. (2020). Survey of Automatic Spelling Correction, Electronics, vol. 9, no. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/electronics9101670\u003c/span\u003e\u003cspan address=\"10.3390/electronics9101670\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJurafsky, D., \u0026amp; James, H. M. (2018). N-gram language models, Speech and language processing, 23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, S., Lee, J., Moon, H., Park, C., Seo, J., Eo, S., Koo, S., \u0026amp; Lim, H. (2023). A Survey on Evaluation Metrics for Machine Translation, Mathematics, vol. 4 (11), no. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/math11041006\u003c/span\u003e\u003cspan address=\"10.3390/math11041006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, p. 1006.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHugging Face, 11 9 2024. [Online]. Available: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://huggingface.co/docs/evaluate/index\u003c/span\u003e\u003cspan address=\"https://huggingface.co/docs/evaluate/index\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. [Accessed 9 11 2024].\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Ottoman-Turkish transliteration, Ottoman-Turkish conversion, Orthographic transliteration, NLP techniques, Dictionary-based transliteration","lastPublishedDoi":"10.21203/rs.3.rs-5735281/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5735281/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eOttoman-Turkish transliteration is a relatively new problem. To make a vast amount of historical documents, books, newspapers, and magazines accessible to a wider audience unfamiliar with the Ottoman script, it is necessary to transliterate the Ottoman script into the Latin-based Turkish script. This study employs traditional NLP techniques to develop a dictionary-based Ottoman-Turkish transliteration system. Using a dataset of 2403 sentences and 31K words, we achieved a Word Error Rate (WER) of 20.69% (raw), 6.31% (normalized) and a Character Error Rate (CER) of 6.46% (raw) 3.01% (normalized), resulting in a BLEU score of 51.90 (raw) 77.18 (normalized). The results show that the proposed system has a promising performance for Ottoman-Turkish transliteration.\u003c/p\u003e","manuscriptTitle":"The Ottoman-Turkish Transliteration using Traditional NLP Techniques","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-02 06:09:58","doi":"10.21203/rs.3.rs-5735281/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4344b555-d834-4136-8ceb-085bef544490","owner":[],"postedDate":"January 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-02-11T15:38:53+00:00","versionOfRecord":[],"versionCreatedAt":"2025-01-02 06:09:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5735281","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5735281","identity":"rs-5735281","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}