Event Detection in Multiple Webpages based on Comprehensive Dimension Matching and Co-occurrence Constraint
Yuanzi Xu, Qingzhong Li, Zhongmin Yan, Wei Wang · Applied Mathematics & Information Sciences · 2014
Abstract: Detecting various sentence-level events from multiple webpages can be important in finding knowledge. We propose an event detection method based on comprehensive dimension matching and co-occurrence constraint. First, we detect events from a single webpage by clustering co-reference sentence-level event mentions. These events are considered as co-occurrence events in every single webpage. Second, similar events from multiple webpages are clustered. The dimension matching method is used to aggregate event mentions. Different matchers measure different dimensions, and an extended evidence theory is proposed to allocate dynamic weight and combine dimension measurement results. We propose an event co-occurrence constraint to reduce match times and quantity of candidate matches events in the multiple webpages event-detection process to improve event cluster efficiency. The experiment results demonstrate that this method can detect various events and noticeably reduce the quantity of co-reference events.