A Survey on the Identification of Causal Relation in Texts
Mingyue Han, Yinglin Wang · 2021
Causality is the basis for humans to make rational decisions and is widely mentioned in different fields. In the natural language processing (NLP) community, the problem of causality is complex and challenging. This paper serves as an effort to briefly discuss the causal relation identification in texts, from the existing causal resources, research methodology, and the robustness problems. First, we introduce relevant causal datasets and resources. Second, the existing typical approaches that have been used in causal relation identification are categorized into unsupervised and supervised methods. In addition, the robustness of causality identification models is discussed succinctly. Finally, we try to list the research challenges at present and raise the future research directions in this field.