KnowDis: Knowledge Enhanced Data Augmentation for Event Causality Detection via Distant Supervision

• DOI: 10.18653/v1/2020.coling-main.135
• Publication Date: 2021-01-08T13:58:31Z
• Authors: Xinyu Zuo; Yubo Chen; Kang Liu; Jun Zhao
• Keywords: Margin (machine learning); Causality; Benchmark (surveying); Labeled data; Training set; Commonsense knowledge

Modern models of event causality detection (ECD) are mainly based on supervised learning from small hand-labeled corpora. However, training data is expensive to produce, low coverage causal expressions, and limited in size, which makes methods hard detect relations between events. To solve this lacking problem, we investigate a augmentation framework for ECD, dubbed as Knowledge Enhanced Distant Data Augmentation (KnowDis). Experimental results two benchmark datasets EventStoryLine corpus Causal-TimeBank show that 1) KnowDis can augment available assisted with the lexical commonsense knowledge ECD via distant supervision, 2) our method outperforms previous by large margin automatically labeled data.