An Overview of Event Extraction Methods based on Semantic Disambiguation
Haili Li, Xiaodong Wang, Yunyan Zhou, Weijie Liu, Shilong Pan · 2024
Event extraction is a fundamental and complex task in information extraction, aiming at automatically identifying and extracting structured event information from unstructured text. However, the same word may have different meanings in different contexts, necessitating semantic disambiguation during the classification process. The polysemy of triggers presents a significant challenge to accurate EE. Existing surveys primarily focus on different domains, fields, or technical paradigms, but none provide a systematic summary of the semantic disambiguation techniques employed. This study classifies these techniques into three categories based on the underlying technical frameworks and offers a comprehensive overview of the most advanced methods in each category. Finally, we summarize the application scenarios, strengths, and limitations of existing approaches, and discuss potential directions for future research.