Next-Gen Data Analytics and Intelligent Automation
Dynamic Text-Attributed Graphs and Learning Models for Community Evolution: A Survey of Recent Advances
Authors: Sruthi K S, P B Divya, A Sreekumar and Kannan Balakrishnan
Publishing Date: 24-12-2025
ISBN: 978-81-975670-6-3
Abstract
Understanding community changes in dynamic networks is a key research area with applications in social networks, academic networks, biological networks, and financial networks. While traditional graph models focus on their structural patterns, real-world systems often combine these patterns with substantial textual content associated with nodes and edges, resulting in Dynamic Text-Attributed Graphs (DyTAGs). Such networks record both time-based interactions and the development of textual details, allowing for a more thorough analysis of how communities are created, changed, and evolved. This work reviews recent progress in dynamic graph learning for community detection in text-attributed networks. This paper discusses the framework for handling Dynamic Text-Attributed Graphs (DyTAGs). It surveys advanced models, including TGAT, DyGFormer, and methods involving Large Language Models (LLMs). The research emphasizes the importance of integrating temporal, structural, and semantic data and shows that this multimodal integration improves both the precision and the understanding of the community evolution of networks.
Keywords
Complex Networks, Dynamic Text Attributed Graphs, Community Evolution, Graph Neural Networks, Large Language Models.
Cite as
Sruthi K S, P B Divya, A Sreekumar and Kannan Balakrishnan, "Dynamic Text-Attributed Graphs and Learning Models for Community Evolution: A Survey of Recent Advances", In: Kusum Kumari Bharti and Noor Firdoos Jahan (eds), Next-Gen Data Analytics and Intelligent Automation, SCRS, India, 2025, pp. 48-62. https://doi.org/10.56155/978-81-975670-6-3-5