رحمن عميري

Mining data from social networks is crucial as it provides deep insights into user behavior, trends, and interactions, which are invaluable for various applications. Consequently, tracking social network community evolutions offers valuable insights into the patterns of community evolution over time. While the cornerstone of community evolution revolves around discussion topics, most studies in this field have focused on complex network structure, overlooking the influence of topics and topical features to increase accuracy and reduce complexity time. An additional challenge stemming from prior tracking methodologies lies in simplifying process of identifying accurate community evolution events within social networks. Addressing this challenge necessitates the development of a more robust methodology. This study develops and proposes three methods to track community evolution. First, by integrating two important fields: topic modeling and network structure, we propose two dynamic models for tracking and predicting the evolution of social communities. The first model is tailored for tracking the evolution of communities centered on a single topic, denoted as the "tracking topical community evolution model." The second model extends this by encompassing communities engaged in discussions spanning multiple topics, denoted as the "tracking multi-topic community evolution model." The third model, “IEGM,” is proposed to enhance the process of tracking and identifying community evolution types. This model considers graph structures in tracking community evolution, simplifying the identification of evolution events with high accuracy and less complexity and time. The suitability of our models for predicting the evolution of community areas was further investigated through a prediction study. In this eva‎luation, three classifiers—RF Classifier, Light GBM, and XGB Classifier—were employed. The results obtained from the proposed models and the IEGM method demonstrate their effectiveness in enhancing the accuracy of tracking and predicting community evolution. Additionally, these models contribute to reducing complexity and consuming less time, thereby facilitating accurate decision-making in a diverse range of applications that require a profound understanding and emphasis on community topics.

پيش‌بيني تكامل جامعه , ويژگي‌هاي ساختاري , ويژگي‌هاي موضوعي , مدل‌سازي موضوع , رديابي تكامل جامعه

Prediction of Community Evolution , structural features , Topical Features , Topic Modelling , Tracking Community Evolution

Rahman Omeiri

Dr. Hassan Naderi