Explicit and implicit knowledge-enhanced model for event causality identification.

Event Causality Identification (ECI) aims at detecting the causal relation between 2 events, which is a challenging task due to the complexity of causal expressions and the background knowledge needed for identifying certain causal relations. Considerable work has been done on the learning of context and the incorporation of external knowledge. However, none of the work incorporates both explicit and implicit causal knowledge. To this end, we propose an integrative model for event causality identification, integrating both explicit causal indicators and implicit causal knowledge with the data-oriented model. For the data-oriented model, an event pair graph is constructed and Relational Graph Convolutional Network (R-GCN) is employed to better capture interactions between individual pairs. Regarding the explicit causal indicators, their word embeddings are used to initialize the filters of convolutional neural network so as to capture the clues indicating causal relation. We further introduce a cause–effect matching mechanism to better leverage implicit causal knowledge. It measures the possibility of causal relation holding between 2 events based on the possible causes and effects generated by COMET. The proposed method is evaluated on 3 datasets, and experimental results demonstrate the effectiveness and superiority of the proposed method. • Propose a novel method integrating RGCN and explicit and implicit causal knowledge. • Propose a cause–effect matching mechanism to leverage implicit causal knowledge. • Achieve the best performance on 3 widely adopted datasets. [ABSTRACT FROM AUTHOR]