This paper addresses the problem of document-level event extraction as a critical step toward automated knowledge acquisition for intelligent systems. Extracting structured event knowledge from unstructured text is essential for downstream knowledge-based applications such as knowledge graph construction, decision support, and intelligent question answering. However, existing approaches struggle to effectively capture long-range dependencies and complex inter-event relationships within documents while maintaining computational efficiency. To address these challenges, we propose a novel knowledge-aware framework, the Graph Convolutional Network with Pseudo-trigger Combination Recognition(GCN-PCR). The proposed model explicitly represents entities, mentions, and sentences as interconnected nodes in a relational graph, enabling structured modeling of semantic dependencies and co-reference information across document contexts. Furthermore, the pseudo-trigger combination strategy enhances the identification of event structures by improving the robustness of event boundary detection and argument association. Extensive experiments on two public datasets demonstrate that the proposed approach achieves superior performance compared to state-of-the-art methods. More importantly, our framework provides a scalable and effective solution for transforming unstructured textual data into structured event knowledge, facilitating its integration into knowledge-based systems.
原文链接
分享
引用
