The advance of event relation extraction based on deep learning

With the development of machine learning and natural language processing, events and event relations, as indispensable elements of communication, have been paid more and more attention. The early use of machine learning for event extraction is based on vocabulary and context features. With the introduction of an inference mechanism and neural network, the performance of event extraction is greatly improved. Complete events contain many internal components (such as people, time, and place) and external relationships (such as timing, cause, and effect). Event relation extraction for many events can train the model to predict and realize a deeper understanding of text information. At the same time, event relationship extraction plays an essential role in many fields such as medical treatment, business, and law. The current event extraction model is mainly based on predicate recognition. However, it is far from achieving the target of complete event recognition. Besides, the accuracy of event extraction and prediction based on multiple-choice is not good. This paper reviews the main methods of event relation extraction. Furthermore, this paper discusses the prospect and challenge of event relation extraction.