BGCFormer: A Text Event Feature Fusion Learning Model based on Transformer

Event extraction is an essential task in natural language processing, as it involves extracting meaningful events from text documents, which is important for a variety of applications, such as information retrieval, question answering, and text summarization, yet challenges remain when extracting events from documents, which is that a document usually contain multiple sentences together form a complete event, and entities in the same event may span multiple sentences. To address these challenges, this paper proposes a transformer with features fusion learning model (BGCFormer), which is based on a transformer architecture with GCN and encoder attention mechanism, and it can build a feature fusion learning network to capture global interaction features between different sentences and entity mentions. Experiments conducted on a large-scale dataset have demonstrated the proposed model outperforms existing methods in terms of accuracy and efficiency.