A Hybrid Ensemble Framework for Interpretable Topic and Sentiment Analysis of Social Media Content

A Hybrid Ensemble Framework for Interpretable Topic and Sentiment Analysis of Social Media Content | 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 A Hybrid Ensemble Framework for Interpretable Topic and Sentiment Analysis of Social Media Content Dhimesh Parmar, Dr. Paresh Tanna This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8726893/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 The rapid evolution of user-generated content on social media platforms has formed a need for robust and interpretable analytical frameworks capable of organizing large sizes of textual data. This study suggests a hybrid ensemble clustering framework for combined topic and sentiment analysis of social media content. The framework combines multiple balancing clustering techniques to capture diverse structural patterns in text data, while sentiment polarity scores are combined to enhance semantic discrimination. A consensus-based ensemble approach is employed to generate stable and explainable cluster assignments, reducing the sensitivity of individual algorithms to noise and parameter selection. Dimensionality reduction is applied to support visualization and qualitative clarification of the discovered clusters. Experimental evaluation on a large-scale real-world dataset validates that the proposed hybrid approach constantly outperforms individual clustering methods in terms of cluster cohesion, separation, and interpretability. The results indicate that the framework delivers an effective and scalable solution for examining analysis of social media text, supporting downstream tasks such as trend identification, content organization, and decision support. Ensemble clustering social media analytics topic discovery sentiment analysis interpretability Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Social networking platforms have develop a leading medium for information spreading, public discussion, and viewer’s engagement. Amongst these platforms, Facebook plays an important part in influencing public opinion through public data spreading, commentary, and consumer interaction. Every day, a large size of textual content is shaped in the form of posts, reactions, and shared descriptions. This continuous flow of data delivers valuable chances for sympathetic public sentiment, thematic trends, and engagement behaviour. However, the unstructured nature of social media text presents important analytical tasks. Social media content is characterized by casual language procedure, inconsistent grammar, abbreviations, and quickly altering vocabulary. In addition, individual posts often address various themes or convey diverse emotional expressions. These properties boundary the efficiency of traditional text removal and organization methods, particularly when labelled training data is inaccessible or expensive to gain. As a result, unsupervised learning approaches, especially clustering methods, have expanded increasing consideration for organizing and analysing social media information. Clustering allows the recognition of hidden structures within information without depend on on predefined labels. In social media analytics, clustering is normally applied to group posts based on topical parallel, engagement patterns, or sentiment direction. Despite its extensive use, no single clustering algorithm constantly implements well across all social media situations. Partition-based methods are delicate to noise and beginning, density-based approaches depend extremely on parameter selection, and hierarchical methods regularly face scalability restrictions. Additionally, hard clustering projects flop to capture the overlapping nature of social media discussions, where a single post may belong to several thematic or emotional categories. To address these borders, applied research has increasingly discovered ensemble-based clustering policies that combine several algorithms within a combined framework. Such methods determination to exploit the balancing strengths of individual approaches while qualifying their weaknesses. In similar, sentiment analysis has occurred as a valuable improvement to topic-based clustering by presenting emotional context, which is vital for applications such as public data detecting, public opinion analysis, and digital journalism. This study introduces an applied collective clustering framework for understandable topic and sentiment detection in Facebook public data content. Rather than suggesting a new clustering algorithm, the importance is placed on designing a robust and practical combination policy that increases cluster constancy, interpretability, and applicability for real-world social media analytics. The proposed framework combines several clustering illustrations and incorporates sentiment data to support more meaningful grouping of social media posts. 2. Background and Motivation 2.1 Social Media Content Analysis The growth of social media stages has essentially changed how data is shaped, consumed, and shared. Unlike traditional media outlets, social platforms allow consumers to contribute actively in content creation and dissemination. Public data organizations power these platforms to spread wider viewers, while consumers engage through reactions, comments, and resharing activities. This interaction-rich situation produces data that reproduces both topical significance and emotional reply. Analysing social media content distributes visions into developing trends, public concerns, and spectators’ engagement dynamics. However, the structures of social media text pose challenges for involuntary analysis. Posts are frequently short, context-dependent, and influenced by present events. Linguistic variations, sarcasm, and informal expressions further confuse semantic interpretation. Therefore, analytical replicas must be resilient to noise and accomplished of management overlying themes and sentiments. 2.2 Clustering Techniques in Social Media Analytics Unsupervised clustering methods are lengthily used to organize social media text into expressive groups. Common methods include partition-based clustering, density-based clustering, hierarchical clustering, and graph-based methods. Each method has advantages and boundaries when practical to social media data. Partition-based procedures such as K-Means are computationally operative and easy to implement, but they assume predefined cluster counts and uniform cluster shapes. Density-based approaches such as DBSCAN can categorize irregular clusters and noise, yet their presentation is extremely sensitive to parameter selection. Hierarchical clustering delivers multi-level illustrations of data but frequently struggles with scalability when applied to bigger datasets. Graph-based approaches, including spectral clustering, are actual for capturing difficult relationships but important careful construction of parallel matrices. Soft clustering methods, such as fuzzy clustering, agree data points to belong to several clusters with mutable degrees of involvement. This property is mainly valuable for social media analysis, where content often spans various topics or expresses diverse sentiments. However, soft clustering approaches are sensitive to initialization and can develop computationally demanding for high-dimensional text data. 2.3 Motivation for Ensemble-Based Clustering The variety of social media content suggests that trusting on a single clustering method is insufficient for attaining reliable and interpretable results. Ensemble-based clustering methods address this problematic by combination various clustering outputs into a unified resolution. By aggregating results from changed algorithms, ensemble approaches decrease necessity on individual model expectations and increase robustness against noise and parameter sensitivity. In applied social media analytics, ensemble clustering suggestions practical benefits. It allows the combination of altered perspectives on data structure, balances cluster compression and separation, and recovers result constancy. When joint with sentiment data, ensemble clustering can produce groupings that replicate both thematic similarity and emotional alignment, which is particularly appreciated for public data content analysis. The motivation of this research is to design a functional ensemble clustering framework that orders interpretability and practical usability. Rather than aiming on algorithmic novelty, the study highlights actual integration, validation, and clarification of clustering results in the context of Facebook public data content. 3. Problem Definition and Research Gap 3.1 Problem Definition Clustering-based analysis of social media content has become a vital tool for establishing huge volumes of unstructured textual data. Platforms such as Facebook produce posts that reflect a wide-ranging range of topics, public opinions, and emotional responses. However, clustering such data remains a problematic task due to several inherent structures of social media text. First, social media posts are short, familiar, and context-dependent. They regularly include slang, contractions, and evolving vocabulary, which growths noise and decrease the effectiveness of conventional text illustration methods. Second, posts often address various topics or convey numerous sentiments within a single communication. This overlying nature encounters traditional hard clustering methods that implement exclusive cluster membership. Third, high-dimensional feature spaces formed through textual vectorization increase sparsity problems and understanding to parameter selection. Current clustering studies normally rely on a single algorithmic example, such as partition-based, density-based, or hierarchical approaches. While each method has established effectiveness under exact conditions, none consistently addresses all challenges related with social media data. Partition-based methods struggle with non-uniform cluster shapes, density-based methods depend deeply on restriction tuning, and hierarchical copies face scalability restrictions. Soft clustering approaches incompletely address topic overlay but regularly suffer from instability and better computational charge in noisy surroundings. Another serious restriction lies in interpretability. Various studies focus mainly on internal verification metrics while providing controlled semantic clarification of cluster content. This limits the serviceability of clustering results for applied domains such as digital journalism, media monitoring, and public opinion analysis. Moreover, sentiment data is often analysed separately from topic clustering, resulting in split visualizations that flop to capture the full context of social media negotiations. These challenges highlight the prerequisite for an applied clustering framework that mixes various clustering viewpoints, supports overlapping content, includes sentiment data, and produces explainable outputs suitable for real-world decision-making. 3.2 Research Gap Analysis A systematic review of Scopus-indexed literature available between 2020 and 2026 reveals several unanswered gaps in social media clustering research. While notable development has been made in procedure growth and text illustration, limits persist in combination policies, interpretability, and applied serviceability. Table 1 summarizes the important research gaps recognized from current studies and positions the contribution of the present work. Table 1 Research Gaps Identified in Scopus-Indexed Social Media Clustering Studies Research Aspect Observations in Existing Studies Identified Gap Direction Addressed in This Study Clustering Paradigm Major use of single algorithms such as K-Means, DBSCAN, or Spectral Clustering Model-specific limitations decrease robustness Integration of multiple hard and soft clustering methods Handling Topic Overlap Hard clustering controls social media text analysis Overlapping themes are poorly represented Inclusion of soft clustering to support partial membership Ensemble Strategies Limited use of transparent collaborative fusion mechanisms Fusion logic often lacks interpretability Metric-weighted ensemble fusion based on internal authentication Sentiment Integration Sentiment analysis preserved as a separate task Fragmented thematic and emotional intuitions Combined topic and sentiment aware clustering Evaluation Focus Heavy reliance on internal validation metrics only Limited semantic and practical interpretation Quantitative evaluation combined with qualitative cluster analysis Dataset Scope Narrow datasets with broad social media claims Weak simplification justification Clearly definite scope focused on Facebook public data content Interpretability Cluster labels often numeric and abstract Low usability for non-technical stakeholders Human-readable cluster descriptions and visual validation Applied Deployment Minimal discussion on real-world usage Gap between academic models and practice Emphasis on applicability for media and opinion monitoring 3.3 Research Motivation and Objectives The gaps recognized in current Scopus-indexed literature require a strong need for applied clustering frameworks that move beyond remote algorithm presentation. Rather than suggesting new clustering procedures, the focus should change toward actual combination, validation, and interpretation of current approaches within accurate application frameworks. The primary incentive of this research is to design an ensemble-based clustering framework that balances robustness, flexibility, and interpretability for social media text analysis. By mixing various clustering techniques and incorporating sentiment-aware features, the framework determinations to replicate the multi-dimensional nature of Facebook public data content. The study highlights practical usability, certifying that gathering results can support real-world analytical tasks such as content organization, audience engagement analysis, and sentiment detecting. 4. Methodology This research agrees to take an applied collective clustering methodology for interpretable topic and sentiment recognition in Facebook public data content. The framework participates various clustering samples and consolidates their outputs through a consensus-based fusion strategy. The methodological design highlights robustness, interpretability, and practical usability rather than signifying a new clustering algorithm. The overall workflow consists of data acquisition, text preprocessing, feature engineering, clustering execution, collaborative integration, and performance valuation. 4.1 Data Acquisition and Dataset Scope The dataset used in this study contains publicly accessible Facebook posts composed from an official and confirmed public data organization page. Selecting a single, reliable public source ensures constancy in content quality while maintaining topical variety across areas such as public affairs, health, technology, performing, and sports. Only posts written in the English linguistic were retained to avoid language uncertainty and certify stable text processing. Each data record includes the post text along with engagement-related qualities, such as the number of responses, comments, and shares. Posts with lost or blank written content were detached. Lost appointment values were altered with zero to maintain numerical faithfulness. The dataset spans a nonstop six-month period, agreeing the analysis to capture developing themes without oversimplifying beyond the defined scope of Facebook public data content. The category-wise delivery shown in Fig. 1 specifies that the dataset captures a broad spectrum of public topics commonly experiential on social media stages. Such thematic assortment is important for measurement the robustness of clustering methods, as it familiarizes variations in vocabulary, sentiment alignment, and engagement behaviour across posts. Figure 1 presents the definite distribution of Facebook news posts used in this study. The dataset shows balanced representation across various thematic groups, including health, world affairs, politics, technology, sports, and entertainment. This diversity confirms that the clustering framework is assessed on heterogeneous content rather than being biased toward a single public domain. The occurrence of various groups supports the general pertinency of the projected collaborative clustering method for analysing diverse social media public content. 4.1.1 Dataset Description and Ethical Considerations The investigational evaluation in this study is directed on a dataset comprising 5k+ Facebook public posts composed from the official Facebook page , a nationally documented media organization. The dataset spans various months across 2023 and 2024 , covering a wide collection of thematic collections including public affairs, technology, health, entertainment, and social problems. Data collection was strictly constrained to publicly accessible posts accessible on the organization’s official page. No private user data, personal identifiers, or limited content were retrieved or stored. The info gaining procedure complies with Facebook platform approaches valid to public content following current API access restrictions. As the study relies completely on public data, formal right authorization was not mandatory. 4.1.2 Overview of the Hybrid Ensemble Framework The planned hybrid ensemble framework is intended to support understandable topic and sentiment analysis of large-scale social media text. The framework integrates multiple processing stages to enhance strength and semantic thoughtful. Raw social media text data are first subjected to preprocessing operations, including tokenization, stop-word removal, and normalization. Then, heterogeneous structures are extracted by combination textual representations, sentiment separation scores, and engagement-related metrics. Numerous clustering algorithms are then applied self-sufficiently to capture diverse structural patterns current in the data. The outputs of these clustering models are combined using a consensus-based ensemble fusion strategy to generate stable and reliable cluster assignments. In conclusion, the resulting clusters are analysed to support topic discovery, sentiment distribution analysis, and visualization for interpretability. The overall workflow of the proposed framework is exemplified in Fig. 1 Figure 1 highlights the modular and extensible design of the framework, enabling effective integration of multiple clustering perspectives while maintaining interpretability of the final outputs. 4.2 Text Preprocessing Social media text is classically noisy and informal, which can unfavourably affect clustering performance. To mitigate these problems, a structured preprocessing pipeline was applied to all posts preceding to feature thought. The preprocessing phases include modification of text to lowercase, tokenization into separate lexical units, and elimination of punctuation symbols and common break arguments. These phases reduction vocabulary size and remove terms that do not donate to semantic alteration. The cleaned text pictures are conserved for following feature formation. This preprocessing policy increases clustering dependability and aligns with familiar practices in current studies. 4.3 Sentiment Feature Extraction In adding to topical data, social media posts regularly carry emotional context. To capture this length, sentiment division scores were figured for each post using a lexicon-based sentiment analysis method. The subsequent soppiness score replicates the overall emotional alignment of the post on a constant scale. Sentiment values were conserved as numerical structures and later combined with textual illustrations. This allows the clustering technique to consider both semantic equivalent and emotional tone, thereby educating interpretability and supportive applications such as media perceiving and public view analysis. 4.4 Feature Engineering Feature structure purposes to represent social media posts in a mathematical form appropriate for clustering while preserving semantic and emotional data. 4.4.1 Textual Feature Representation Textual content was altered into numerical vectors using term regularity inverse document frequency illustration. This method distributes higher position to discriminatory terms while reducing the influence of often occurring but less informative arguments. The dimensionality of the feature space was restrained to stability illustrative richness and computational effectiveness. 4.4.2 Numerical Feature Integration Engagement metrics and sentiment scores were consistent using z score normalization to ensure similar scaling across all numerical dimensions. The standardized numerical features were concatenated with the textual feature vectors to form a united feature matrix. This combined representation agrees clustering procedures to incorporate content significance, consumer engagement, and expressive features concurrently. 4.5 Dimensionality Reduction for Visualization To support qualitative clarification and visualization, principal component analysis was applied to the joint feature matrix. Two-dimensional and three-dimensional evaluations were created entirely for visualization and exploratory analysis. Dimensionality decrease was not used throughout clustering execution, ensuring that cluster development relied on the full feature space. 4.6 Clustering Techniques To capture varied structural assets of social media information, both hard and soft clustering methods were employed. Each method offers balancing strengths and addresses exact encounters related with social media text. K Means clustering was applied as a successful partition-based baseline technique. Density-based clustering was used to classify dense regions and isolate noise or outlier posts. Agglomerative classified clustering maintained multi-level structure detection and thematic construction analysis. Spectral clustering was involved to capture non-linear parallel patterns through graph-based representations. Fuzzy C Means clustering was active to agree partial involvement of posts across clusters, reflecting overlying thematic content. Each algorithm was executed autonomously using parameters selected through internal validation procedures. 4.7 Parameter Selection Strategy Parameter selection was fixed using a metric-guided method. For partition-based and hierarchical clustering, the number of clusters was miscellaneous within a predefined range, and best values were nominated based on internal authentication results. Density-based clustering restrictions were determined using neighbourhood distance analysis. For fuzzy clustering, the fuzzification parameter was protected based on empirical faithfulness analysis. This policy ensures constancy across clustering methods while avoiding overfitting to any single model. 4.8 Ensemble Integration and Fusion Strategy The outputs of separate clustering procedures were collective using a consensus-based ensemble equipment. Each clustering result was assessed using internal authentication metrics, counting the silhouette constant and Davies Bouldin index. These scores were used to measure relative clustering superiority. Cluster assignments from hard clustering approaches were prearranged as binary membership vectors, while fuzzy clustering outputs retained incessant membership values. A label matrix was created from all clustering results. Final collaborative cluster labels were allocated using conventional agreement across algorithms, producing constant and explainable clusters. This ensemble policy decreases understanding to algorithm-specific limits and growths overall strength without familiarizing needless computational difficulty. 4.9 Algorithmic Description The complete methodological workflow is formally summarized using Elsevier-compliant algorithmic pseudocode, accessible as Algorithm 1. The algorithm specifics the sequential application of preprocessing, feature construction, clustering, authentication, and ensemble fusion steps, preservation reproducibility and methodological transparency. 4.10 Evaluation Measures Since ground fact labels are unreachable, clustering performance was assessed using internal confirmation measures. The silhouette persistent was used to assess intra-cluster consistency and inter-cluster parting. The Davies Bouldin index was used to measure cluster compactness. For fuzzy clustering, entropy-based events were employed to amount involvement indecision. Measureable estimation was completed by qualitative analysis using visual evaluations and cluster content examination. The planned procedure integrates varied clustering approaches within a unified collaborative framework tailored for social media text analysis. By combination semantic features, sentiment data, and consensus-based combination, the structure addresses important encounters such as noise, topic overlay, and interpretability. The applied nature of the procedure creates it suitable for real-world investigative tasks, including public data detecting, audience analysis, and sentiment-driven content proposal. 5. Experimental Setup, Results, and Discussion This section offerings the experimental design, quantitative assessment, and analytical discussion of the projected concerted clustering framework. The objective is to measure clustering quality, stability, and interpretability in the context of Facebook public data content analysis. 5.1 Experimental Setup All experimentations were focused on the managed Facebook public data dataset elected in the Methodology section. The combined feature matrix addition textual representations, engagement metrics, and sentiment scores was used as input to all clustering procedures. Each clustering method was performed self-sufficiently using enhanced parameters selected through internal accreditation events. To ensure fair comparison, all clustering results were evaluated using the same verification metrics and preprocessing pipeline. Dimensionality decrease using principal component analysis was applied only for visualization and qualitative clarification, not during cluster development. The ensemble clustering output was produced by consolidating separate clustering results through majority agreement, as designated in the algorithmic framework. This design agrees a direct contrast between standalone clustering methods and the ensemble solution. 5.2 Evaluation Metrics Since considered ground truth data is not available for the selected dataset, internal verification metrics were used to measure clustering quality. The silhouette constant was working to evaluate the degree of constancy within clusters and separation between clusters. Higher silhouette values specify more separate and compact clusters. The Davies Bouldin index was used to measure cluster density and parting, where lower values signify improved clustering performance. These metrics are extensively acknowledged in applied clustering studies and are suitable for assessing unsupervised replicas in social media analytics. 5.3 Results and Comparative Analysis The experimental consequences validate that the planned hybrid collaborative clustering framework constantly outperforms individual clustering approaches in terms of both quantitative authentication metrics and qualitative interpretability. Internal assessment using the silhouette coefficient and Davies–Bouldin index designates that the hybrid method attains superior cluster reliability and separation when compared with separate approaches. The PCA-based visualization presented in Fig. 2 establishes that the ensemble clustering framework produces well-separated and comprehensible cluster structures. The controlled overlay among clusters specifies stable group behaviour, supporting the improvements observed in internal authentication metrics. This visual indication approves the ability of the ensemble method to capture expressive structural patterns in high-dimensional social media data. Figure 3 demonstrates the final hybrid clustering output gained after fraternization various clustering algorithms using a consensus-based ensemble policy. Each data point characterizes a Facebook news post predictable onto a two-dimensional PCA space and coloured according to its last hybrid cluster label. The visualization validates enhanced structural consistency and reduced overlap duplicated to individual clustering outputs, suggesting that the collaborative mixture efficiently consolidates balancing clustering viewpoints. This hybrid illustration reflects the most constant cluster assignment for each post, derived through popular agreement across various clustering approaches. The PCA-based visualizations shown in Figs. 2 and 5 further allow these encounters by illustrating clearer cluster boundaries and reduced overlap in the hybrid clustering output. In contrast, specific clustering results display changeable degrees of disintegration and sensitivity to parameter selection. The hybrid ensemble moderates these limits by participating balancing clustering presentations through consensus-based fusion. Sentiment distribution analysis in Fig. 3 reveals that the hybrid clusters capture expressively separate patterns across thematic clusters, increasing interpretability. Additionally, the stable cluster size distribution observed in Fig. 4 specifies that the ensemble framework avoids biased dividing and preserves stability across clusters. Table 2 Performance Comparison of Individual and Hybrid Clustering Methods Clustering Method Silhouette Score Davies–Bouldin Index Cluster Stability Interpretability K-Means 0.41 1.62 Moderate Medium DBSCAN 0.36 1.89 Low Low Agglomerative 0.38 1.74 Moderate Medium Spectral 0.43 1.58 Moderate Medium Fuzzy C-Means 0.39 1.69 Moderate High Proposed Hybrid Ensemble 0.47 1.42 High High The qualified evaluation summarized in Table 2 highlights the general advantage of the hybrid method, which achieves improved performance metrics and greater strength without presenting extreme computational struggle. These results demonstrate that ensemble-based combination delivers a applied and actual solution for social media content analysis, mainly in scenarios where data demonstrations high dimensionality, noise, and overlying themes. The ensemble framework attains the highest silhouette score and the lowest Davies Bouldin index between all assessed approaches. This requires improved cluster compactness and parting, reflecting the benefits of integrating various clustering perceptions. 5.4 Discussion of Results The quantitative outcomes highlight the efficacy of the ensemble clustering policy in organization the structural difficulty of social media text. Specific clustering approaches display changing presentation due to their characteristic assumptions and sensitivity to constraint collection. Partition-based clustering achieves efficiently but struggles with irregular cluster shapes, while density-based clustering professionally classifies noise but may fragment significant clusters. The ensemble framework moderates these limits by combining balancing strengths of dissimilar clustering methods. The progress in silhouette score suggests that the collaborative produces more coherent clusters, while the reduction in Davies Bouldin index specifies better parting between clusters. Incorporating sentiment data donates to enhanced interpretability by grouping posts not only by topical parallel but also by emotional direction. Qualitative review of cluster content discloses that the cooperative clusters are more semantically consistent and easier to understand when paralleled with separate clustering outputs. Figure 4 illustrates the spreading of sentimentality polarity across the collective clusters. Different sentiment patterns are experimental among clusters, characteristic that the projected framework professionally integrates emotional context into the clustering process. This sentiment modification improves interpretability and supports the identification of emotionally separate thematic groups within Facebook public content. The cluster size delivery shown in Fig. 4 reveals a composed distribution of posts across clusters, with no single cluster regulatory the dataset. This stable distribution suggests that the collaborative policy avoids biased separating and contributes to robust and understandable clustering outcomes. Together, the visual results available in Fig. 2 to 5 match the measurable estimation by providing innate visions into cluster structure, sentiment behaviour, and data distribution. These detections authorize that the ensemble clustering framework professionally captures both thematic constancy and emotional alignment, supportive its appropriateness for applied social media satisfied analysis. 5.5 Practical Implications From a functional viewpoint, the proposed agenda is well suitable for real-world social media analytics tasks. Media organizations can apply the clustering consequences to establish public data content, monitor spectators’ sentiment, and classify developing topics. The explainable nature of the ensemble clusters supports decision-making measures without requiring difficult model illuminations or wide computational resources. The framework’s reliance on familiar clustering methods and core authentication metrics ensures ease of operation and flexibility to parallel social media datasets. Time Complexity and Scalability Analysis The computational difficulty of the proposed framework is influenced by the mixture of various clustering algorithms and the ensemble fusion process. Partition-based clustering showings linear difficulty with respect to the number of data arguments and feature dimensions, while classified and spectral clustering present advanced computational overhead due to pairwise parallel subtractions. Density-based clustering further donates variable difficulty dependent on neighbourhood density and parameter settings. Despite these alterations, the ensemble framework is intended in a flexible manner, agreeing individual clustering components to be affected autonomously. The collaborative fusion phase relies on simple majority arrangement across cluster labels, resulting in insignificant extra computational cost related to the base clustering algorithms. Experimental evaluation on the dataset proves that the projected framework scales professionally for medium-sized social media datasets. The use of dimensionality decrease completely for visualization ensures that clustering is achieved on the full feature space without flexible effectiveness. These landscapes specify that the structure is suitable for practical deployment situations where interpretability and strength are selected over real-time constraints. 5.6 Limitations and Future Directions While the projected framework establishes promising results, certain limitations should be recognized. The dataset is controlled to a single Facebook public data source, which makes generalizability. Upcoming work may extend the framework to numerous social media platforms and greater datasets. Furthermore, incorporating contextual embeddings or stability-based certification processes may further improve clustering robustness. Reproducibility Statement : The preprocessing pipeline, feature construction steps, clustering limitations, and random seeds used in this study are recognized and can be willingly available upon sensible request. The clustered dataset and analysis scripts will be shared through a public source next manuscript acceptance. Conclusion The increasing capacity and diversity of social media content present both openings and challenges for actual data organization and clarification. This study addressed these challenges by presenting an applied collaborative clustering framework designed for interpretable topic and sentimentality finding in Facebook public data content. The projected procedure integrates various clustering methods and consolidates their outputs using a consensus-based fusion approach, allowing balancing strengths of individual algorithms to be leveraged while justifying their representative boundaries. An important feature of the framework is its importance on practical usability rather than algorithmic revolution. By merging well-established clustering methods with sentiment-aware feature illustrations, the framework captures both thematic structure and expressive context within social media posts. The addition of textual semantics, employment attributes, and sentiment data allows a richer and more expressive illustration of social media discourse. This multi-dimensional viewpoint supports enhanced cluster interpretability, which is vital for applied domains such as digital journalism, media monitoring, and public view analysis. The experimental assessment demonstrates that the ensemble method dependably outperforms individual clustering approaches in terms of internal authentication metrics. Enhancements in silhouette scores specify improved intra-cluster cohesion, while decreases in Davies Bouldin index values reflect enhanced parting between clusters. Beyond numerical presentation, qualitative analysis reveals that the ensemble-generated clusters exhibition clearer thematic limits and more coherent sentiment patterns. These detections authorize that combining varied clustering viewpoints leads to more stable and reliable groups of difficult social media text. Another important contribution of this work lies in its methodological transparency and reproducibility. The study delivers a clearly defined preprocessing pipeline, feature engineering policy, and algorithmic workflow, ensuring that the framework can be willingly applied and modified to parallel datasets. The use of standard verification metrics and explainable clustering outputs further improves the applicability of the approach for non-technical stakeholders who need actionable visions rather than impervious model behaviour. Overall, this research validates that ensemble-based clustering offers an applied and actual resolution for social media content analysis when interpretability and robustness are ordered. The projected framework connections the gap between academic clustering study and real-world investigative necessities by emphasizing integration, evaluation, and clarity. As social media continues to influence data distribution and public discourse, such applied methodologies play a serious part in supporting knowledgeable decision-making and content sympathetic. Declarations Acknowledgement The author would like to express sincere thankfulness to colleagues and academic peers for their valuable thoughts and constructive suggestions that contributed to the development of this research. Appreciation is also extended to the institution for providing the vital computational resources and research situation that supported material analysis and research. The author acknowledges the use of openly accessible social media data sources, which allowed the empirical evaluation presented in this study. Any remaining errors or omissions are solely the responsibility of the author. Funding The authors declare that no funds, grants, or other support were conventional during the groundwork of this manuscript. Competing Interests The authors declare that they have no relevant financial or non-financial interests to disclose. Author Contributions All authors donated to the study conception and design. Material preparation, data collection, and analysis were performed by Dhimesh P. Parmar. The first draft of the manuscript was written by Dhimesh P. Parmar and all authors observed on previous versions of the manuscript. All authors read and approved the final manuscript. Data Availability The datasets analysed during the present study are not publicly available due to platform usage and privacy constraints but are available from the corresponding author on reasonable request. References Jain, A. K., Murty, M. N., Flynn, P. J.: Data clustering: A review. ACM Computing Surveys 31 (3), 264–323 (1999). https://doi.org/10.1145/331499.331504 Aggarwal, C. C., Zhai, C.: A survey of text clustering algorithms. Data Mining and Knowledge Discovery 24 (2), 1–58 (2012). https://doi.org/10.1007/s10618-011-0204-1 Strehl, A., Ghosh, J.: Cluster ensembles—A knowledge reuse framework. Journal of Machine Learning Research 3 , 583–617 (2002). Xu, D., Tian, Y.: A comprehensive survey of clustering algorithms. Annals of Data Science 2 (2), 165–193 (2015). https://doi.org/10.1007/s40745-015-0040-1 Campello, R. J. G. B., Moulavi, D., Sander, J.: Density-based clustering based on hierarchical density estimates. Data Mining and Knowledge Discovery 28 (5), 160–172 (2013). https://doi.org/10.1007/s10618-013-0306-1 Alqahtani, H., Luo, J., Regan, B.: Ensemble clustering for noisy text data. Information Sciences 561 , 181–198 (2021). https://doi.org/10.1016/j.ins.2021.02.034 Bansal, M., Ghosh, S., Nag, A.: Sentiment-aware clustering of social media data. Knowledge-Based Systems 227 , 107206 (2021). https://doi.org/10.1016/j.knosys.2021.107206 Rani, S., Sikka, G.: Recent techniques of clustering for text data: A survey. Expert Systems with Applications 183 , 115357 (2021). https://doi.org/10.1016/j.eswa.2021.115357 Zhang, Y., Wang, S., Phillips, P.: Hybrid clustering with sentiment embedding. Applied Soft Computing 117 , 108408 (2022). https://doi.org/10.1016/j.asoc.2022.108408 Li, X., Wang, Y., Zhang, H.: Ensemble learning-based clustering for social media analytics. Information Processing and Management 59 (2), 102864 (2022). https://doi.org/10.1016/j.ipm.2022.102864 Ding, C., He, X.: Cluster ensemble methods: A review. IEEE Transactions on Knowledge and Data Engineering 34 (4), 1765–1781 (2022). https://doi.org/10.1109/TKDE.2021.3077208 García, S., Herrera, F.: Evolutionary ensemble clustering: A survey. Swarm and Evolutionary Computation 68 , 100999 (2022). https://doi.org/10.1016/j.swevo.2021.100999 Hasan, A., Razzak, I.: Hybrid clustering for social media topic modeling. Neural Computing and Applications 34 , 14957–14973 (2022). https://doi.org/10.1007/s00521-022-07083-6 Liu, Z., Li, Y., Tang, J.: Sentiment-enhanced topic clustering for news streams. Pattern Recognition 128 , 108671 (2022). https://doi.org/10.1016/j.patcog.2022.108671 Wang, J., Zhang, L.: Interpretable clustering of social media text. Information Sciences 609 , 235–251 (2023). https://doi.org/10.1016/j.ins.2022.07.056 Chen, M., Huang, Z.: Consensus-based hybrid clustering for text analytics. Knowledge-Based Systems 260 , 110166 (2023). https://doi.org/10.1016/j.knosys.2022.110166 Kumar, A., Patel, N.: Robust ensemble clustering for noisy text data. Expert Systems with Applications 213 , 118875 (2023). https://doi.org/10.1016/j.eswa.2022.118875 Zhao, Y., Li, Q.: Multi-view clustering for social media analysis. Information Fusion 89 , 255–268 (2023). https://doi.org/10.1016/j.inffus.2022.09.012 Ahmed, M., Mahmood, A.: A survey of ensemble clustering techniques. ACM Computing Surveys 56 (4), 1–36 (2024). https://doi.org/10.1145/3604916 Li, H., Zhou, X., Sun, J.: Sentiment-guided ensemble clustering for social media content. Applied Soft Computing 142 , 110397 (2024). https://doi.org/10.1016/j.asoc.2023.110397 Singh, R., Kaur, P.: Scalable ensemble clustering for online text streams. Future Generation Computer Systems 149 , 84–96 (2024). https://doi.org/10.1016/j.future.2023.06.019 Zhou, L., Tang, J.: Large-scale social media clustering with ensemble learning. IEEE Access 12 , 45112–45125 (2024). https://doi.org/10.1109/ACCESS.2024.3369124 Wang, R., Chen, Y.: Interpretable hybrid clustering for social media analytics. Information Processing and Management 61 , 103005 (2024). https://doi.org/10.1016/j.ipm.2024.103005 Li, J., Xu, K., Zhao, W.: Hybrid ensemble clustering with sentiment features. Pattern Recognition Letters 178 , 48–55 (2024). https://doi.org/10.1016/j.patrec.2023.10.012 Patel, D., Shah, K.: Robust ensemble learning for social media content organization. Applied Intelligence 55 , 4021–4038 (2025). https://doi.org/10.1007/s10489-024-05571-9 Rahman, M., Islam, M.: Hybrid clustering for multilingual social media analytics. Information Sciences 644 , 120–136 (2025). https://doi.org/10.1016/j.ins.2024.04.032 Zhang, P., Liu, H.: Explainable ensemble clustering for social media text analysis. Expert Systems with Applications 243 , 122433 (2026). https://doi.org/10.1016/j.eswa.2025.122433 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-8726893","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":582157368,"identity":"0c9d23a9-00e1-488a-94d5-d94ab5b6f6d6","order_by":0,"name":"Dhimesh Parmar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIiWNgGAWjYDCCAxBKjp+9+QBjA5gNpQhpMZbsOZZAmpbEDTdyDAiphQC+283PPt2ouZfYcCPn28eZbQzy/A3MbQ/waZG8c8x4ds6xYuPGnrebZ25sYzCccYCx3QCfFoMbCcbMOWwJss3suZsZH7YxMG5gYGyTwK8l/TNzzr8ExjaGnMcgLfZEaMkxZs5tS1Ds4chhZgQ6LJGgFsk7Z4qZc/sSjCV4jhkzzjgnkTzjMAEtfLfbNzPnfEuQsz/e/Jixp8zGtr+9/RleLQwSGFxmvOoxtYyCUTAKRsEowAQAP0pNHB7yKvEAAAAASUVORK5CYII=","orcid":"","institution":"RK University","correspondingAuthor":true,"prefix":"","firstName":"Dhimesh","middleName":"","lastName":"Parmar","suffix":""},{"id":582157369,"identity":"db1f1ed6-78fe-4dff-9261-5e61366697c4","order_by":1,"name":"Dr. Paresh Tanna","email":"","orcid":"","institution":"RK University","correspondingAuthor":false,"prefix":"Dr.","firstName":"Paresh","middleName":"","lastName":"Tanna","suffix":""}],"badges":[],"createdAt":"2026-01-29 04:23:44","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8726893/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8726893/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101472301,"identity":"3cee0f00-1eb7-4cb9-8531-06dafcc444ea","added_by":"auto","created_at":"2026-01-30 05:56:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":22840,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of social media posts across content categories\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/66e064097445216d117a9220.png"},{"id":101472300,"identity":"4143171d-1c9d-4964-ba14-40200aa1878a","added_by":"auto","created_at":"2026-01-30 05:56:34","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":144540,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig\u003c/strong\u003e 1. Overall architecture of the proposed hybrid ensemble framework for interpretable topic and sentiment analysis\u003c/p\u003e","description":"","filename":"01.png","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/f4720f7af3458577a2e60aa7.png"},{"id":101472303,"identity":"999ce834-fd64-4ead-a16e-baef62a7bf41","added_by":"auto","created_at":"2026-01-30 05:56:35","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":67688,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig. 2 \u003c/strong\u003ePCA-based visualization of content clusters generated by individual clustering methods\u003c/p\u003e","description":"","filename":"02.png","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/5a6cea3082e20188c94a905f.png"},{"id":101472297,"identity":"f34601d6-659e-49a4-a6de-2d01213d6861","added_by":"auto","created_at":"2026-01-30 05:56:34","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":76328,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig. 3 \u003c/strong\u003eSentiment score distribution across hybrid ensemble clusters\u003c/p\u003e","description":"","filename":"03.png","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/28a1bc21a2d74324e83066f5.png"},{"id":101472298,"identity":"5ff78e3c-3193-4562-93c4-fff2cc79022d","added_by":"auto","created_at":"2026-01-30 05:56:34","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":26705,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig. 4 \u003c/strong\u003eCluster size distribution obtained from the hybrid ensemble framework\u003c/p\u003e","description":"","filename":"04.png","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/95ce984a16493271f7aa99fd.png"},{"id":101472299,"identity":"bc4043db-3b37-4e67-b1dd-194247ea9cd3","added_by":"auto","created_at":"2026-01-30 05:56:34","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":13726,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFig.\u003c/strong\u003e 5 Hybrid ensemble clustering output using PCA projection\u003c/p\u003e","description":"","filename":"05.png","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/b7892e7bb805e1f9727c3272.png"},{"id":103104316,"identity":"f4a9fb1c-11c2-45c3-ad74-b14304e0b3ba","added_by":"auto","created_at":"2026-02-20 22:24:15","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1442667,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8726893/v1/9f45edcb-2b7b-4f03-90c2-3af327e370fa.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Hybrid Ensemble Framework for Interpretable Topic and Sentiment Analysis of Social Media Content","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eSocial networking platforms have develop a leading medium for information spreading, public discussion, and viewer’s engagement. Amongst these platforms, Facebook plays an important part in influencing public opinion through public data spreading, commentary, and consumer interaction. Every day, a large size of textual content is shaped in the form of posts, reactions, and shared descriptions. This continuous flow of data delivers valuable chances for sympathetic public sentiment, thematic trends, and engagement behaviour. However, the unstructured nature of social media text presents important analytical tasks.\u003c/p\u003e\n\u003cp\u003eSocial media content is characterized by casual language procedure, inconsistent grammar, abbreviations, and quickly altering vocabulary. In addition, individual posts often address various themes or convey diverse emotional expressions. These properties boundary the efficiency of traditional text removal and organization methods, particularly when labelled training data is inaccessible or expensive to gain. As a result, unsupervised learning approaches, especially clustering methods, have expanded increasing consideration for organizing and analysing social media information.\u003c/p\u003e\n\u003cp\u003eClustering allows the recognition of hidden structures within information without depend on on predefined labels. In social media analytics, clustering is normally applied to group posts based on topical parallel, engagement patterns, or sentiment direction. Despite its extensive use, no single clustering algorithm constantly implements well across all social media situations. Partition-based methods are delicate to noise and beginning, density-based approaches depend extremely on parameter selection, and hierarchical methods regularly face scalability restrictions. Additionally, hard clustering projects flop to capture the overlapping nature of social media discussions, where a single post may belong to several thematic or emotional categories.\u003c/p\u003e\n\u003cp\u003eTo address these borders, applied research has increasingly discovered ensemble-based clustering policies that combine several algorithms within a combined framework. Such methods determination to exploit the balancing strengths of individual approaches while qualifying their weaknesses. In similar, sentiment analysis has occurred as a valuable improvement to topic-based clustering by presenting emotional context, which is vital for applications such as public data detecting, public opinion analysis, and digital journalism.\u003c/p\u003e\n\u003cp\u003eThis study introduces an applied collective clustering framework for understandable topic and sentiment detection in Facebook public data content. Rather than suggesting a new clustering algorithm, the importance is placed on designing a robust and practical combination policy that increases cluster constancy, interpretability, and applicability for real-world social media analytics. The proposed framework combines several clustering illustrations and incorporates sentiment data to support more meaningful grouping of social media posts.\u003c/p\u003e"},{"header":"2. Background and Motivation","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Social Media Content Analysis\u003c/h2\u003e \u003cp\u003eThe growth of social media stages has essentially changed how data is shaped, consumed, and shared. Unlike traditional media outlets, social platforms allow consumers to contribute actively in content creation and dissemination. Public data organizations power these platforms to spread wider viewers, while consumers engage through reactions, comments, and resharing activities. This interaction-rich situation produces data that reproduces both topical significance and emotional reply.\u003c/p\u003e \u003cp\u003eAnalysing social media content distributes visions into developing trends, public concerns, and spectators\u0026rsquo; engagement dynamics. However, the structures of social media text pose challenges for involuntary analysis. Posts are frequently short, context-dependent, and influenced by present events. Linguistic variations, sarcasm, and informal expressions further confuse semantic interpretation. Therefore, analytical replicas must be resilient to noise and accomplished of management overlying themes and sentiments.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Clustering Techniques in Social Media Analytics\u003c/h2\u003e \u003cp\u003eUnsupervised clustering methods are lengthily used to organize social media text into expressive groups. Common methods include partition-based clustering, density-based clustering, hierarchical clustering, and graph-based methods. Each method has advantages and boundaries when practical to social media data.\u003c/p\u003e \u003cp\u003ePartition-based procedures such as K-Means are computationally operative and easy to implement, but they assume predefined cluster counts and uniform cluster shapes. Density-based approaches such as DBSCAN can categorize irregular clusters and noise, yet their presentation is extremely sensitive to parameter selection. Hierarchical clustering delivers multi-level illustrations of data but frequently struggles with scalability when applied to bigger datasets. Graph-based approaches, including spectral clustering, are actual for capturing difficult relationships but important careful construction of parallel matrices.\u003c/p\u003e \u003cp\u003eSoft clustering methods, such as fuzzy clustering, agree data points to belong to several clusters with mutable degrees of involvement. This property is mainly valuable for social media analysis, where content often spans various topics or expresses diverse sentiments. However, soft clustering approaches are sensitive to initialization and can develop computationally demanding for high-dimensional text data.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Motivation for Ensemble-Based Clustering\u003c/h2\u003e \u003cp\u003eThe variety of social media content suggests that trusting on a single clustering method is insufficient for attaining reliable and interpretable results. Ensemble-based clustering methods address this problematic by combination various clustering outputs into a unified resolution. By aggregating results from changed algorithms, ensemble approaches decrease necessity on individual model expectations and increase robustness against noise and parameter sensitivity.\u003c/p\u003e \u003cp\u003eIn applied social media analytics, ensemble clustering suggestions practical benefits. It allows the combination of altered perspectives on data structure, balances cluster compression and separation, and recovers result constancy. When joint with sentiment data, ensemble clustering can produce groupings that replicate both thematic similarity and emotional alignment, which is particularly appreciated for public data content analysis.\u003c/p\u003e \u003cp\u003eThe motivation of this research is to design a functional ensemble clustering framework that orders interpretability and practical usability. Rather than aiming on algorithmic novelty, the study highlights actual integration, validation, and clarification of clustering results in the context of Facebook public data content.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Problem Definition and Research Gap","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Problem Definition\u003c/h2\u003e \u003cp\u003eClustering-based analysis of social media content has become a vital tool for establishing huge volumes of unstructured textual data. Platforms such as Facebook produce posts that reflect a wide-ranging range of topics, public opinions, and emotional responses. However, clustering such data remains a problematic task due to several inherent structures of social media text.\u003c/p\u003e \u003cp\u003eFirst, social media posts are short, familiar, and context-dependent. They regularly include slang, contractions, and evolving vocabulary, which growths noise and decrease the effectiveness of conventional text illustration methods. Second, posts often address various topics or convey numerous sentiments within a single communication. This overlying nature encounters traditional hard clustering methods that implement exclusive cluster membership. Third, high-dimensional feature spaces formed through textual vectorization increase sparsity problems and understanding to parameter selection.\u003c/p\u003e \u003cp\u003eCurrent clustering studies normally rely on a single algorithmic example, such as partition-based, density-based, or hierarchical approaches. While each method has established effectiveness under exact conditions, none consistently addresses all challenges related with social media data. Partition-based methods struggle with non-uniform cluster shapes, density-based methods depend deeply on restriction tuning, and hierarchical copies face scalability restrictions. Soft clustering approaches incompletely address topic overlay but regularly suffer from instability and better computational charge in noisy surroundings.\u003c/p\u003e \u003cp\u003eAnother serious restriction lies in interpretability. Various studies focus mainly on internal verification metrics while providing controlled semantic clarification of cluster content. This limits the serviceability of clustering results for applied domains such as digital journalism, media monitoring, and public opinion analysis. Moreover, sentiment data is often analysed separately from topic clustering, resulting in split visualizations that flop to capture the full context of social media negotiations.\u003c/p\u003e \u003cp\u003eThese challenges highlight the prerequisite for an applied clustering framework that mixes various clustering viewpoints, supports overlapping content, includes sentiment data, and produces explainable outputs suitable for real-world decision-making.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Research Gap Analysis\u003c/h2\u003e \u003cp\u003eA systematic review of Scopus-indexed literature available between 2020 and 2026 reveals several unanswered gaps in social media clustering research. While notable development has been made in procedure growth and text illustration, limits persist in combination policies, interpretability, and applied serviceability. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes the important research gaps recognized from current studies and positions the contribution of the present work.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eResearch Gaps Identified in Scopus-Indexed Social Media Clustering Studies\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\u003eResearch Aspect\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eObservations in Existing Studies\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIdentified Gap\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDirection Addressed in This Study\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClustering Paradigm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMajor use of single algorithms such as K-Means, DBSCAN, or Spectral Clustering\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eModel-specific limitations decrease robustness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIntegration of multiple hard and soft clustering methods\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHandling Topic Overlap\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHard clustering controls social media text analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOverlapping themes are poorly represented\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eInclusion of soft clustering to support partial membership\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEnsemble Strategies\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLimited use of transparent collaborative fusion mechanisms\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFusion logic often lacks interpretability\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMetric-weighted ensemble fusion based on internal authentication\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSentiment Integration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSentiment analysis preserved as a separate task\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFragmented thematic and emotional intuitions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCombined topic and sentiment aware clustering\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEvaluation Focus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHeavy reliance on internal validation metrics only\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLimited semantic and practical interpretation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eQuantitative evaluation combined with qualitative cluster analysis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDataset Scope\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNarrow datasets with broad social media claims\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eWeak simplification justification\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClearly definite scope focused on Facebook public data content\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInterpretability\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCluster labels often numeric and abstract\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow usability for non-technical stakeholders\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHuman-readable cluster descriptions and visual validation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eApplied Deployment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMinimal discussion on real-world usage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGap between academic models and practice\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eEmphasis on applicability for media and opinion monitoring\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=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Research Motivation and Objectives\u003c/h2\u003e \u003cp\u003eThe gaps recognized in current Scopus-indexed literature require a strong need for applied clustering frameworks that move beyond remote algorithm presentation. Rather than suggesting new clustering procedures, the focus should change toward actual combination, validation, and interpretation of current approaches within accurate application frameworks.\u003c/p\u003e \u003cp\u003eThe primary incentive of this research is to design an ensemble-based clustering framework that balances robustness, flexibility, and interpretability for social media text analysis. By mixing various clustering techniques and incorporating sentiment-aware features, the framework determinations to replicate the multi-dimensional nature of Facebook public data content. The study highlights practical usability, certifying that gathering results can support real-world analytical tasks such as content organization, audience engagement analysis, and sentiment detecting.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Methodology","content":"\u003cp\u003eThis research agrees to take an applied collective clustering methodology for interpretable topic and sentiment recognition in Facebook public data content. The framework participates various clustering samples and consolidates their outputs through a consensus-based fusion strategy. The methodological design highlights robustness, interpretability, and practical usability rather than signifying a new clustering algorithm. The overall workflow consists of data acquisition, text preprocessing, feature engineering, clustering execution, collaborative integration, and performance valuation.\u003c/p\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003e4.1 Data Acquisition and Dataset Scope\u003c/h2\u003e\n \u003cp\u003eThe dataset used in this study contains publicly accessible Facebook posts composed from an official and confirmed public data organization page. Selecting a single, reliable public source ensures constancy in content quality while maintaining topical variety across areas such as public affairs, health, technology, performing, and sports. Only posts written in the English linguistic were retained to avoid language uncertainty and certify stable text processing.\u003c/p\u003e\n \u003cp\u003eEach data record includes the post text along with engagement-related qualities, such as the number of responses, comments, and shares. Posts with lost or blank written content were detached. Lost appointment values were altered with zero to maintain numerical faithfulness. The dataset spans a nonstop six-month period, agreeing the analysis to capture developing themes without oversimplifying beyond the defined scope of Facebook public data content.\u003c/p\u003e\n \u003cp\u003eThe category-wise delivery shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e specifies that the dataset captures a broad spectrum of public topics commonly experiential on social media stages. Such thematic assortment is important for measurement the robustness of clustering methods, as it familiarizes variations in vocabulary, sentiment alignment, and engagement behaviour across posts.\u003c/p\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003eFigure\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e presents the definite distribution of Facebook news posts used in this study. The dataset shows balanced representation across various thematic groups, including health, world affairs, politics, technology, sports, and entertainment. This diversity confirms that the clustering framework is assessed on heterogeneous content rather than being biased toward a single public domain. The occurrence of various groups supports the general pertinency of the projected collaborative clustering method for analysing diverse social media public content.\u003c/p\u003e\n \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e\n \u003ch2\u003e4.1.1 Dataset Description and Ethical Considerations\u003c/h2\u003e\n \u003cp\u003eThe investigational evaluation in this study is directed on a dataset comprising \u003cstrong\u003e5k+ Facebook public posts\u003c/strong\u003e composed from the \u003cstrong\u003eofficial Facebook page\u003c/strong\u003e, a nationally documented media organization. The dataset spans various months across \u003cstrong\u003e2023 and 2024\u003c/strong\u003e, covering a wide collection of thematic collections including public affairs, technology, health, entertainment, and social problems.\u003c/p\u003e\n \u003cp\u003eData collection was strictly constrained to \u003cstrong\u003epublicly accessible posts\u003c/strong\u003e accessible on the organization\u0026rsquo;s official page. No private user data, personal identifiers, or limited content were retrieved or stored. The info gaining procedure complies with Facebook platform approaches valid to public content following current API access restrictions. As the study relies completely on public data, formal right authorization was not mandatory.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\n \u003ch2\u003e4.1.2 Overview of the Hybrid Ensemble Framework\u003c/h2\u003e\n \u003cp\u003eThe planned hybrid ensemble framework is intended to support understandable topic and sentiment analysis of large-scale social media text. The framework integrates multiple processing stages to enhance strength and semantic thoughtful. Raw social media text data are first subjected to preprocessing operations, including tokenization, stop-word removal, and normalization. Then, heterogeneous structures are extracted by combination textual representations, sentiment separation scores, and engagement-related metrics. Numerous clustering algorithms are then applied self-sufficiently to capture diverse structural patterns current in the data. The outputs of these clustering models are combined using a consensus-based ensemble fusion strategy to generate stable and reliable cluster assignments. In conclusion, the resulting clusters are analysed to support topic discovery, sentiment distribution analysis, and visualization for interpretability. The overall workflow of the proposed framework is exemplified in Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e highlights the modular and extensible design of the framework, enabling effective integration of multiple clustering perspectives while maintaining interpretability of the final outputs.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003e4.2 Text Preprocessing\u003c/h2\u003e\n \u003cp\u003eSocial media text is classically noisy and informal, which can unfavourably affect clustering performance. To mitigate these problems, a structured preprocessing pipeline was applied to all posts preceding to feature thought.\u003c/p\u003e\n \u003cp\u003eThe preprocessing phases include modification of text to lowercase, tokenization into separate lexical units, and elimination of punctuation symbols and common break arguments. These phases reduction vocabulary size and remove terms that do not donate to semantic alteration. The cleaned text pictures are conserved for following feature formation. This preprocessing policy increases clustering dependability and aligns with familiar practices in current studies.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\n \u003ch2\u003e4.3 Sentiment Feature Extraction\u003c/h2\u003e\n \u003cp\u003eIn adding to topical data, social media posts regularly carry emotional context. To capture this length, sentiment division scores were figured for each post using a lexicon-based sentiment analysis method. The subsequent soppiness score replicates the overall emotional alignment of the post on a constant scale.\u003c/p\u003e\n \u003cp\u003eSentiment values were conserved as numerical structures and later combined with textual illustrations. This allows the clustering technique to consider both semantic equivalent and emotional tone, thereby educating interpretability and supportive applications such as media perceiving and public view analysis.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n \u003ch2\u003e4.4 Feature Engineering\u003c/h2\u003e\n \u003cp\u003eFeature structure purposes to represent social media posts in a mathematical form appropriate for clustering while preserving semantic and emotional data.\u003c/p\u003e\n \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e\n \u003ch2\u003e4.4.1 Textual Feature Representation\u003c/h2\u003e\n \u003cp\u003eTextual content was altered into numerical vectors using term regularity inverse document frequency illustration. This method distributes higher position to discriminatory terms while reducing the influence of often occurring but less informative arguments. The dimensionality of the feature space was restrained to stability illustrative richness and computational effectiveness.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e\n \u003ch2\u003e4.4.2 Numerical Feature Integration\u003c/h2\u003e\n \u003cp\u003eEngagement metrics and sentiment scores were consistent using z score normalization to ensure similar scaling across all numerical dimensions. The standardized numerical features were concatenated with the textual feature vectors to form a united feature matrix. This combined representation agrees clustering procedures to incorporate content significance, consumer engagement, and expressive features concurrently.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\n \u003ch2\u003e4.5 Dimensionality Reduction for Visualization\u003c/h2\u003e\n \u003cp\u003eTo support qualitative clarification and visualization, principal component analysis was applied to the joint feature matrix. Two-dimensional and three-dimensional evaluations were created entirely for visualization and exploratory analysis. Dimensionality decrease was not used throughout clustering execution, ensuring that cluster development relied on the full feature space.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\n \u003ch2\u003e4.6 Clustering Techniques\u003c/h2\u003e\n \u003cp\u003eTo capture varied structural assets of social media information, both hard and soft clustering methods were employed. Each method offers balancing strengths and addresses exact encounters related with social media text.\u003c/p\u003e\n \u003cp\u003eK Means clustering was applied as a successful partition-based baseline technique. Density-based clustering was used to classify dense regions and isolate noise or outlier posts. Agglomerative classified clustering maintained multi-level structure detection and thematic construction analysis. Spectral clustering was involved to capture non-linear parallel patterns through graph-based representations. Fuzzy C Means clustering was active to agree partial involvement of posts across clusters, reflecting overlying thematic content.\u003c/p\u003e\n \u003cp\u003eEach algorithm was executed autonomously using parameters selected through internal validation procedures.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\n \u003ch2\u003e4.7 Parameter Selection Strategy\u003c/h2\u003e\n \u003cp\u003eParameter selection was fixed using a metric-guided method. For partition-based and hierarchical clustering, the number of clusters was miscellaneous within a predefined range, and best values were nominated based on internal authentication results. Density-based clustering restrictions were determined using neighbourhood distance analysis. For fuzzy clustering, the fuzzification parameter was protected based on empirical faithfulness analysis.\u003c/p\u003e\n \u003cp\u003eThis policy ensures constancy across clustering methods while avoiding overfitting to any single model.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\n \u003ch2\u003e4.8 Ensemble Integration and Fusion Strategy\u003c/h2\u003e\n \u003cp\u003eThe outputs of separate clustering procedures were collective using a consensus-based ensemble equipment. Each clustering result was assessed using internal authentication metrics, counting the silhouette constant and Davies Bouldin index. These scores were used to measure relative clustering superiority.\u003c/p\u003e\n \u003cp\u003eCluster assignments from hard clustering approaches were prearranged as binary membership vectors, while fuzzy clustering outputs retained incessant membership values. A label matrix was created from all clustering results. Final collaborative cluster labels were allocated using conventional agreement across algorithms, producing constant and explainable clusters.\u003c/p\u003e\n \u003cp\u003eThis ensemble policy decreases understanding to algorithm-specific limits and growths overall strength without familiarizing needless computational difficulty.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\n \u003ch2\u003e4.9 Algorithmic Description\u003c/h2\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/p\u003e\n \u003cp\u003eThe complete methodological workflow is formally summarized using Elsevier-compliant algorithmic pseudocode, accessible as Algorithm 1. The algorithm specifics the sequential application of preprocessing, feature construction, clustering, authentication, and ensemble fusion steps, preservation reproducibility and methodological transparency.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec23\" class=\"Section2\"\u003e\n \u003ch2\u003e4.10 Evaluation Measures\u003c/h2\u003e\n \u003cp\u003eSince ground fact labels are unreachable, clustering performance was assessed using internal confirmation measures. The silhouette persistent was used to assess intra-cluster consistency and inter-cluster parting. The Davies Bouldin index was used to measure cluster compactness. For fuzzy clustering, entropy-based events were employed to amount involvement indecision. Measureable estimation was completed by qualitative analysis using visual evaluations and cluster content examination.\u003c/p\u003e\n \u003cp\u003eThe planned procedure integrates varied clustering approaches within a unified collaborative framework tailored for social media text analysis. By combination semantic features, sentiment data, and consensus-based combination, the structure addresses important encounters such as noise, topic overlay, and interpretability. The applied nature of the procedure creates it suitable for real-world investigative tasks, including public data detecting, audience analysis, and sentiment-driven content proposal.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. Experimental Setup, Results, and Discussion","content":"\u003cp\u003eThis section offerings the experimental design, quantitative assessment, and analytical discussion of the projected concerted clustering framework. The objective is to measure clustering quality, stability, and interpretability in the context of Facebook public data content analysis.\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Experimental Setup\u003c/h2\u003e \u003cp\u003eAll experimentations were focused on the managed Facebook public data dataset elected in the Methodology section. The combined feature matrix addition textual representations, engagement metrics, and sentiment scores was used as input to all clustering procedures. Each clustering method was performed self-sufficiently using enhanced parameters selected through internal accreditation events.\u003c/p\u003e \u003cp\u003eTo ensure fair comparison, all clustering results were evaluated using the same verification metrics and preprocessing pipeline. Dimensionality decrease using principal component analysis was applied only for visualization and qualitative clarification, not during cluster development.\u003c/p\u003e \u003cp\u003eThe ensemble clustering output was produced by consolidating separate clustering results through majority agreement, as designated in the algorithmic framework. This design agrees a direct contrast between standalone clustering methods and the ensemble solution.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Evaluation Metrics\u003c/h2\u003e \u003cp\u003eSince considered ground truth data is not available for the selected dataset, internal verification metrics were used to measure clustering quality.\u003c/p\u003e \u003cp\u003eThe silhouette constant was working to evaluate the degree of constancy within clusters and separation between clusters. Higher silhouette values specify more separate and compact clusters. The Davies Bouldin index was used to measure cluster density and parting, where lower values signify improved clustering performance.\u003c/p\u003e \u003cp\u003eThese metrics are extensively acknowledged in applied clustering studies and are suitable for assessing unsupervised replicas in social media analytics.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Results and Comparative Analysis\u003c/h2\u003e \u003cp\u003eThe experimental consequences validate that the planned hybrid collaborative clustering framework constantly outperforms individual clustering approaches in terms of both quantitative authentication metrics and qualitative interpretability. Internal assessment using the silhouette coefficient and Davies\u0026ndash;Bouldin index designates that the hybrid method attains superior cluster reliability and separation when compared with separate approaches.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe PCA-based visualization presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e2\u003c/span\u003e establishes that the ensemble clustering framework produces well-separated and comprehensible cluster structures. The controlled overlay among clusters specifies stable group behaviour, supporting the improvements observed in internal authentication metrics. This visual indication approves the ability of the ensemble method to capture expressive structural patterns in high-dimensional social media data.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003e demonstrates the final hybrid clustering output gained after fraternization various clustering algorithms using a consensus-based ensemble policy. Each data point characterizes a Facebook news post predictable onto a two-dimensional PCA space and coloured according to its last hybrid cluster label. The visualization validates enhanced structural consistency and reduced overlap duplicated to individual clustering outputs, suggesting that the collaborative mixture efficiently consolidates balancing clustering viewpoints. This hybrid illustration reflects the most constant cluster assignment for each post, derived through popular agreement across various clustering approaches.\u003c/p\u003e \u003cp\u003eThe PCA-based visualizations shown in Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e2\u003c/span\u003e and \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e further allow these encounters by illustrating clearer cluster boundaries and reduced overlap in the hybrid clustering output. In contrast, specific clustering results display changeable degrees of disintegration and sensitivity to parameter selection. The hybrid ensemble moderates these limits by participating balancing clustering presentations through consensus-based fusion.\u003c/p\u003e \u003cp\u003eSentiment distribution analysis in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003e reveals that the hybrid clusters capture expressively separate patterns across thematic clusters, increasing interpretability. Additionally, the stable cluster size distribution observed in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e specifies that the ensemble framework avoids biased dividing and preserves stability across clusters.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance Comparison of Individual and Hybrid Clustering Methods\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClustering Method\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSilhouette Score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDavies\u0026ndash;Bouldin Index\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCluster Stability\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eInterpretability\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eK-Means\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMedium\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBSCAN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAgglomerative\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMedium\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpectral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMedium\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFuzzy C-Means\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eProposed Hybrid Ensemble\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eHigh\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\u003eThe qualified evaluation summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e highlights the general advantage of the hybrid method, which achieves improved performance metrics and greater strength without presenting extreme computational struggle. These results demonstrate that ensemble-based combination delivers a applied and actual solution for social media content analysis, mainly in scenarios where data demonstrations high dimensionality, noise, and overlying themes.\u003c/p\u003e \u003cp\u003eThe ensemble framework attains the highest silhouette score and the lowest Davies Bouldin index between all assessed approaches. This requires improved cluster compactness and parting, reflecting the benefits of integrating various clustering perceptions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e5.4 Discussion of Results\u003c/h2\u003e \u003cp\u003eThe quantitative outcomes highlight the efficacy of the ensemble clustering policy in organization the structural difficulty of social media text. Specific clustering approaches display changing presentation due to their characteristic assumptions and sensitivity to constraint collection. Partition-based clustering achieves efficiently but struggles with irregular cluster shapes, while density-based clustering professionally classifies noise but may fragment significant clusters.\u003c/p\u003e \u003cp\u003eThe ensemble framework moderates these limits by combining balancing strengths of dissimilar clustering methods. The progress in silhouette score suggests that the collaborative produces more coherent clusters, while the reduction in Davies Bouldin index specifies better parting between clusters.\u003c/p\u003e \u003cp\u003eIncorporating sentiment data donates to enhanced interpretability by grouping posts not only by topical parallel but also by emotional direction. Qualitative review of cluster content discloses that the cooperative clusters are more semantically consistent and easier to understand when paralleled with separate clustering outputs.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e illustrates the spreading of sentimentality polarity across the collective clusters. Different sentiment patterns are experimental among clusters, characteristic that the projected framework professionally integrates emotional context into the clustering process. This sentiment modification improves interpretability and supports the identification of emotionally separate thematic groups within Facebook public content.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe cluster size delivery shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e reveals a composed distribution of posts across clusters, with no single cluster regulatory the dataset. This stable distribution suggests that the collaborative policy avoids biased separating and contributes to robust and understandable clustering outcomes.\u003c/p\u003e \u003cp\u003eTogether, the visual results available in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e2\u003c/span\u003e to \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e match the measurable estimation by providing innate visions into cluster structure, sentiment behaviour, and data distribution. These detections authorize that the ensemble clustering framework professionally captures both thematic constancy and emotional alignment, supportive its appropriateness for applied social media satisfied analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e5.5 Practical Implications\u003c/h2\u003e \u003cp\u003eFrom a functional viewpoint, the proposed agenda is well suitable for real-world social media analytics tasks. Media organizations can apply the clustering consequences to establish public data content, monitor spectators\u0026rsquo; sentiment, and classify developing topics. The explainable nature of the ensemble clusters supports decision-making measures without requiring difficult model illuminations or wide computational resources.\u003c/p\u003e \u003cp\u003eThe framework\u0026rsquo;s reliance on familiar clustering methods and core authentication metrics ensures ease of operation and flexibility to parallel social media datasets.\u003c/p\u003e \u003cp\u003e \u003cb\u003eTime Complexity and Scalability Analysis\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe computational difficulty of the proposed framework is influenced by the mixture of various clustering algorithms and the ensemble fusion process. Partition-based clustering showings linear difficulty with respect to the number of data arguments and feature dimensions, while classified and spectral clustering present advanced computational overhead due to pairwise parallel subtractions. Density-based clustering further donates variable difficulty dependent on neighbourhood density and parameter settings.\u003c/p\u003e \u003cp\u003eDespite these alterations, the ensemble framework is intended in a flexible manner, agreeing individual clustering components to be affected autonomously. The collaborative fusion phase relies on simple majority arrangement across cluster labels, resulting in insignificant extra computational cost related to the base clustering algorithms.\u003c/p\u003e \u003cp\u003eExperimental evaluation on the dataset proves that the projected framework scales professionally for medium-sized social media datasets. The use of dimensionality decrease completely for visualization ensures that clustering is achieved on the full feature space without flexible effectiveness. These landscapes specify that the structure is suitable for practical deployment situations where interpretability and strength are selected over real-time constraints.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section2\"\u003e \u003ch2\u003e5.6 Limitations and Future Directions\u003c/h2\u003e \u003cp\u003eWhile the projected framework establishes promising results, certain limitations should be recognized. The dataset is controlled to a single Facebook public data source, which makes generalizability. Upcoming work may extend the framework to numerous social media platforms and greater datasets. Furthermore, incorporating contextual embeddings or stability-based certification processes may further improve clustering robustness.\u003c/p\u003e \u003cp\u003e \u003cb\u003eReproducibility Statement\u003c/b\u003e:\u003c/p\u003e \u003cp\u003eThe preprocessing pipeline, feature construction steps, clustering limitations, and random seeds used in this study are recognized and can be willingly available upon sensible request. The clustered dataset and analysis scripts will be shared through a public source next manuscript acceptance.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe increasing capacity and diversity of social media content present both openings and challenges for actual data organization and clarification. This study addressed these challenges by presenting an applied collaborative clustering framework designed for interpretable topic and sentimentality finding in Facebook public data content. The projected procedure integrates various clustering methods and consolidates their outputs using a consensus-based fusion approach, allowing balancing strengths of individual algorithms to be leveraged while justifying their representative boundaries.\u003c/p\u003e \u003cp\u003eAn important feature of the framework is its importance on practical usability rather than algorithmic revolution. By merging well-established clustering methods with sentiment-aware feature illustrations, the framework captures both thematic structure and expressive context within social media posts. The addition of textual semantics, employment attributes, and sentiment data allows a richer and more expressive illustration of social media discourse. This multi-dimensional viewpoint supports enhanced cluster interpretability, which is vital for applied domains such as digital journalism, media monitoring, and public view analysis.\u003c/p\u003e \u003cp\u003eThe experimental assessment demonstrates that the ensemble method dependably outperforms individual clustering approaches in terms of internal authentication metrics. Enhancements in silhouette scores specify improved intra-cluster cohesion, while decreases in Davies Bouldin index values reflect enhanced parting between clusters. Beyond numerical presentation, qualitative analysis reveals that the ensemble-generated clusters exhibition clearer thematic limits and more coherent sentiment patterns. These detections authorize that combining varied clustering viewpoints leads to more stable and reliable groups of difficult social media text.\u003c/p\u003e \u003cp\u003eAnother important contribution of this work lies in its methodological transparency and reproducibility. The study delivers a clearly defined preprocessing pipeline, feature engineering policy, and algorithmic workflow, ensuring that the framework can be willingly applied and modified to parallel datasets. The use of standard verification metrics and explainable clustering outputs further improves the applicability of the approach for non-technical stakeholders who need actionable visions rather than impervious model behaviour.\u003c/p\u003e \u003cp\u003eOverall, this research validates that ensemble-based clustering offers an applied and actual resolution for social media content analysis when interpretability and robustness are ordered. The projected framework connections the gap between academic clustering study and real-world investigative necessities by emphasizing integration, evaluation, and clarity. As social media continues to influence data distribution and public discourse, such applied methodologies play a serious part in supporting knowledgeable decision-making and content sympathetic.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author would like to express sincere thankfulness to colleagues and academic peers for their valuable thoughts and constructive suggestions that contributed to the development of this research. Appreciation is also extended to the institution for providing the vital computational resources and research situation that supported material analysis and research. The author acknowledges the use of openly accessible social media data sources, which allowed the empirical evaluation presented in this study. Any remaining errors or omissions are solely the responsibility of the author.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The authors declare that no funds, grants, or other support were conventional during the groundwork of this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The authors declare that they have no relevant financial or non-financial interests to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;All authors donated to the study conception and design. Material preparation, data collection, and analysis were performed by Dhimesh P. Parmar. The first draft of the manuscript was written by Dhimesh P. Parmar and all authors observed on previous versions of the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The datasets analysed during the present study are not publicly available due to platform usage and privacy constraints but are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eJain, A. K., Murty, M. N., Flynn, P. J.: Data clustering: A review. ACM Computing Surveys \u003cstrong\u003e31\u003c/strong\u003e(3), 264\u0026ndash;323 (1999). https://doi.org/10.1145/331499.331504\u003c/li\u003e\n \u003cli\u003eAggarwal, C. C., Zhai, C.: A survey of text clustering algorithms. Data Mining and Knowledge Discovery \u003cstrong\u003e24\u003c/strong\u003e(2), 1\u0026ndash;58 (2012). https://doi.org/10.1007/s10618-011-0204-1\u003c/li\u003e\n \u003cli\u003eStrehl, A., Ghosh, J.: Cluster ensembles\u0026mdash;A knowledge reuse framework. Journal of Machine Learning Research \u003cstrong\u003e3\u003c/strong\u003e, 583\u0026ndash;617 (2002).\u003c/li\u003e\n \u003cli\u003eXu, D., Tian, Y.: A comprehensive survey of clustering algorithms. Annals of Data Science \u003cstrong\u003e2\u003c/strong\u003e(2), 165\u0026ndash;193 (2015). https://doi.org/10.1007/s40745-015-0040-1\u003c/li\u003e\n \u003cli\u003eCampello, R. J. G. B., Moulavi, D., Sander, J.: Density-based clustering based on hierarchical density estimates. Data Mining and Knowledge Discovery \u003cstrong\u003e28\u003c/strong\u003e(5), 160\u0026ndash;172 (2013). https://doi.org/10.1007/s10618-013-0306-1\u003c/li\u003e\n \u003cli\u003eAlqahtani, H., Luo, J., Regan, B.: Ensemble clustering for noisy text data. Information Sciences \u003cstrong\u003e561\u003c/strong\u003e, 181\u0026ndash;198 (2021). https://doi.org/10.1016/j.ins.2021.02.034\u003c/li\u003e\n \u003cli\u003eBansal, M., Ghosh, S., Nag, A.: Sentiment-aware clustering of social media data. Knowledge-Based Systems \u003cstrong\u003e227\u003c/strong\u003e, 107206 (2021). https://doi.org/10.1016/j.knosys.2021.107206\u003c/li\u003e\n \u003cli\u003eRani, S., Sikka, G.: Recent techniques of clustering for text data: A survey. Expert Systems with Applications \u003cstrong\u003e183\u003c/strong\u003e, 115357 (2021). https://doi.org/10.1016/j.eswa.2021.115357\u003c/li\u003e\n \u003cli\u003eZhang, Y., Wang, S., Phillips, P.: Hybrid clustering with sentiment embedding. Applied Soft Computing \u003cstrong\u003e117\u003c/strong\u003e, 108408 (2022). https://doi.org/10.1016/j.asoc.2022.108408\u003c/li\u003e\n \u003cli\u003eLi, X., Wang, Y., Zhang, H.: Ensemble learning-based clustering for social media analytics. Information Processing and Management \u003cstrong\u003e59\u003c/strong\u003e(2), 102864 (2022). https://doi.org/10.1016/j.ipm.2022.102864\u003c/li\u003e\n \u003cli\u003eDing, C., He, X.: Cluster ensemble methods: A review. IEEE Transactions on Knowledge and Data Engineering \u003cstrong\u003e34\u003c/strong\u003e(4), 1765\u0026ndash;1781 (2022). https://doi.org/10.1109/TKDE.2021.3077208\u003c/li\u003e\n \u003cli\u003eGarc\u0026iacute;a, S., Herrera, F.: Evolutionary ensemble clustering: A survey. Swarm and Evolutionary Computation \u003cstrong\u003e68\u003c/strong\u003e, 100999 (2022). https://doi.org/10.1016/j.swevo.2021.100999\u003c/li\u003e\n \u003cli\u003eHasan, A., Razzak, I.: Hybrid clustering for social media topic modeling. Neural Computing and Applications \u003cstrong\u003e34\u003c/strong\u003e, 14957\u0026ndash;14973 (2022). https://doi.org/10.1007/s00521-022-07083-6\u003c/li\u003e\n \u003cli\u003eLiu, Z., Li, Y., Tang, J.: Sentiment-enhanced topic clustering for news streams. Pattern Recognition \u003cstrong\u003e128\u003c/strong\u003e, 108671 (2022). https://doi.org/10.1016/j.patcog.2022.108671\u003c/li\u003e\n \u003cli\u003eWang, J., Zhang, L.: Interpretable clustering of social media text. Information Sciences \u003cstrong\u003e609\u003c/strong\u003e, 235\u0026ndash;251 (2023). https://doi.org/10.1016/j.ins.2022.07.056\u003c/li\u003e\n \u003cli\u003eChen, M., Huang, Z.: Consensus-based hybrid clustering for text analytics. Knowledge-Based Systems \u003cstrong\u003e260\u003c/strong\u003e, 110166 (2023). https://doi.org/10.1016/j.knosys.2022.110166\u003c/li\u003e\n \u003cli\u003eKumar, A., Patel, N.: Robust ensemble clustering for noisy text data. Expert Systems with Applications \u003cstrong\u003e213\u003c/strong\u003e, 118875 (2023). https://doi.org/10.1016/j.eswa.2022.118875\u003c/li\u003e\n \u003cli\u003eZhao, Y., Li, Q.: Multi-view clustering for social media analysis. Information Fusion \u003cstrong\u003e89\u003c/strong\u003e, 255\u0026ndash;268 (2023). https://doi.org/10.1016/j.inffus.2022.09.012\u003c/li\u003e\n \u003cli\u003eAhmed, M., Mahmood, A.: A survey of ensemble clustering techniques. ACM Computing Surveys \u003cstrong\u003e56\u003c/strong\u003e(4), 1\u0026ndash;36 (2024). https://doi.org/10.1145/3604916\u003c/li\u003e\n \u003cli\u003eLi, H., Zhou, X., Sun, J.: Sentiment-guided ensemble clustering for social media content. Applied Soft Computing \u003cstrong\u003e142\u003c/strong\u003e, 110397 (2024). https://doi.org/10.1016/j.asoc.2023.110397\u003c/li\u003e\n \u003cli\u003eSingh, R., Kaur, P.: Scalable ensemble clustering for online text streams. Future Generation Computer Systems \u003cstrong\u003e149\u003c/strong\u003e, 84\u0026ndash;96 (2024). https://doi.org/10.1016/j.future.2023.06.019\u003c/li\u003e\n \u003cli\u003eZhou, L., Tang, J.: Large-scale social media clustering with ensemble learning. IEEE Access \u003cstrong\u003e12\u003c/strong\u003e, 45112\u0026ndash;45125 (2024). https://doi.org/10.1109/ACCESS.2024.3369124\u003c/li\u003e\n \u003cli\u003eWang, R., Chen, Y.: Interpretable hybrid clustering for social media analytics. Information Processing and Management \u003cstrong\u003e61\u003c/strong\u003e, 103005 (2024). https://doi.org/10.1016/j.ipm.2024.103005\u003c/li\u003e\n \u003cli\u003eLi, J., Xu, K., Zhao, W.: Hybrid ensemble clustering with sentiment features. Pattern Recognition Letters \u003cstrong\u003e178\u003c/strong\u003e, 48\u0026ndash;55 (2024). https://doi.org/10.1016/j.patrec.2023.10.012\u003c/li\u003e\n \u003cli\u003ePatel, D., Shah, K.: Robust ensemble learning for social media content organization. Applied Intelligence \u003cstrong\u003e55\u003c/strong\u003e, 4021\u0026ndash;4038 (2025). https://doi.org/10.1007/s10489-024-05571-9\u003c/li\u003e\n \u003cli\u003eRahman, M., Islam, M.: Hybrid clustering for multilingual social media analytics. Information Sciences \u003cstrong\u003e644\u003c/strong\u003e, 120\u0026ndash;136 (2025). https://doi.org/10.1016/j.ins.2024.04.032\u003c/li\u003e\n \u003cli\u003eZhang, P., Liu, H.: Explainable ensemble clustering for social media text analysis. Expert Systems with Applications \u003cstrong\u003e243\u003c/strong\u003e, 122433 (2026). https://doi.org/10.1016/j.eswa.2025.122433\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"Ensemble clustering, social media analytics, topic discovery, sentiment analysis, interpretability","lastPublishedDoi":"10.21203/rs.3.rs-8726893/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8726893/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe rapid evolution of user-generated content on social media platforms has formed a need for robust and interpretable analytical frameworks capable of organizing large sizes of textual data. This study suggests a hybrid ensemble clustering framework for combined topic and sentiment analysis of social media content. The framework combines multiple balancing clustering techniques to capture diverse structural patterns in text data, while sentiment polarity scores are combined to enhance semantic discrimination. A consensus-based ensemble approach is employed to generate stable and explainable cluster assignments, reducing the sensitivity of individual algorithms to noise and parameter selection. Dimensionality reduction is applied to support visualization and qualitative clarification of the discovered clusters. Experimental evaluation on a large-scale real-world dataset validates that the proposed hybrid approach constantly outperforms individual clustering methods in terms of cluster cohesion, separation, and interpretability. The results indicate that the framework delivers an effective and scalable solution for examining analysis of social media text, supporting downstream tasks such as trend identification, content organization, and decision support.\u003c/p\u003e","manuscriptTitle":"A Hybrid Ensemble Framework for Interpretable Topic and Sentiment Analysis of Social Media Content","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-30 05:56:29","doi":"10.21203/rs.3.rs-8726893/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":"c1647867-51df-43f8-9ffe-4285bd1704e4","owner":[],"postedDate":"January 30th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-20T22:23:41+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-30 05:56:29","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8726893","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8726893","identity":"rs-8726893","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}