Document Level Event Extraction

Document-level event extraction (DEE) aims to automatically identify and structure events and their arguments from entire documents, a task more complex than sentence-level extraction due to challenges like scattered arguments and multiple events within a single document. Current research focuses on improving Large Language Model (LLM) performance through refined prompt engineering, data augmentation techniques (like abstractive summarization), and novel model architectures such as those incorporating relation modeling, multi-channel prediction, or human-like reading processes. These advancements are crucial for improving information extraction from large text corpora, with applications ranging from news analysis and financial reporting to scientific literature review and knowledge graph construction.