Graph Transformer Networks with Syntactic and Semantic Structures for Event Argument Extraction

The goal of Event Argument Extraction (EAE) is to find the role of each\nentity mention for a given event trigger word. It has been shown in the\nprevious works that the syntactic structures of the sentences are helpful for\nthe deep learning models for EAE. However, a major problem in such prior works\nis that they fail to exploit the semantic structures of the sentences to induce\neffective representations for EAE. Consequently, in this work, we propose a\nnovel model for EAE that exploits both syntactic and semantic structures of the\nsentences with the Graph Transformer Networks (GTNs) to learn more effective\nsentence structures for EAE. In addition, we introduce a novel inductive bias\nbased on information bottleneck to improve generalization of the EAE models.\nExtensive experiments are performed to demonstrate the benefits of the proposed\nmodel, leading to state-of-the-art performance for EAE on standard datasets.\n