AI-Augmented Community Sensing for Real-Time Social Influence Mapping

Recent advances in AI (Social-LLM by Jiang & Ferrara, TSGAN by Mane et al.) showcase the power of combining language models with network data, but most applications focus on retrospective analysis or specific user behavior (like aggression). This idea proposes a real-time system that ingests streaming social media data, continuously updating influence maps using a fusion of graph neural networks and LLMs—capturing both network structure and evolving content. The system would detect not only established influencers, but also emerging micro-communities, rising hashtags, and shifting topics in near real-time. The novelty lies in real-time, AI-driven, content-and-structure integration for ongoing influence surveillance, rather than static or user-specific predictions. This approach could revolutionize how researchers, journalists, and organizations track information flows and emerging leaders, providing early warning for viral events, grassroots movements, or coordinated campaigns. It directly builds on Jiang & Ferrara (2023), extending their work toward dynamic, real-world applications.

References:

1. Social-LLM: Modeling User Behavior at Scale using Language Models and Social Network Data. Julie Jiang, Emilio Ferrara (2023). arXiv.org.
2. TSGAN: Temporal Social Graph Attention Network for Aggressive Behavior Forecasting. Swapnil S. Mane, Suman Kundu, Rajesh Sharma (2025). AAAI Conference on Artificial Intelligence.