ناطق الساعدي

In the era of rapidly evolving online communities and social networks, understanding topic evolution is crucial for applications such as trend prediction, community management, and information dissemination. However, existing approaches often emphasize structure dynamics at the expense of the content of user interaction networks, which through metrics such as node centrality, clustering, and community composition play an important role in shaping topic evolution. This study proposes a comprehensive framework for tracking and predicting topic evolution by integrating content discussions and network structural features. Through a series of research steps, including data preprocessing, content topics were identified using advanced topic modeling techniques. Data were divided into “25-time windows” for temporal analysis, and dynamic graphic networks were generated that capture user interaction. Structural features were extracted to track topic evolution, and predictive models were developed to predict future topic trends based on the interaction between content and network structure. We used a method to track the temporal evolution of topics in time series analysis. By integrating the fields of topic modeling and network structure features, we used and eva‎luated “11 machine learning models”. The results indicate that the “CatBoost" model outperforms the other models, achieving superior performance in accuracy (92.6%), F1 scores (92.90), and execution time (0.0211). These models contribute to reducing computational complexity and processing time, thus facilitating more efficient and informed decision-making in applications that require a comprehensive understanding of the dynamic nature of social topics.

رديابي تكامل موضوع , پيش بيني تكامل موضوع , ويژگي هاي ساختار شبكه , تحليل شبكه هاي اجتماعي , مدل سازي موضوع

Tracking Topic evolution , predicting of Topic evolution , Network structure features , social network analysis , Topic Modeling

Nategh Alsaedi

Dr. Naderi