Modeling correlated causal-effect structure with a hypergraph for document-level event causality identification

Wei Xiang, Cheng Liu, Bang Wang

Abstract: Highlights•Propose an event-hypergraph neural encoding model for document-level ECI.•Propose an event semantic encoding module to initialize events’ embeddings.•Propose an event-hypergraph module to model causal-effect correlations.•Experiments on ESC and MECI corpus validate the effectiveness of our EHNEM.