Identify Event Causality with Knowledge and Analogy

Event causality identification (ECI) aims to identify the causal relationship between events, which plays a crucial role in deep text understanding. Due diversity of real-world events and difficulty obtaining sufficient training data, existing ECI approaches have poor generalizability struggle relation seldom seen events. In this paper, we propose utilize both external knowledge internal analogy improve ECI. On one hand, commonsense graph called ConceptNet enrich description an event sample reveal commonalities or associations different other retrieve similar as exam- ples glean useful experiences from such analogous neigh- bors better new pair. By understanding through exter- nal making with can alleviate data sparsity issue model gener- alizability. Extensive evaluations on two benchmark datasets show that our outperforms baseline methods by around 18% F1-value average