EventConnector: Mining Social Event Relations through Temporal Graphs
Researchers have developed EventConnector, a novel framework for understanding and retrieving related real-world events by modeling their temporal dynamics. This system constructs a temporal event graph that captures localized co-fluctuations and lead-lag relationships between events. An adaptive retrieval mechanism, EC-Fusion, further enhances this by fusing the graph-based scores with a Granger-causal signal. Evaluations on prediction market benchmarks and forecasting architectures demonstrated EC-Fusion's superior performance, significantly reducing RMSE compared to existing baselines. AI