摘要

Event duplication detection, also named event coreference finding, is an important subtask of the finding and research of Web entity traces. Based on the ground of market intelligence field, a certain subject can attend only one activity at the same time and same location, and we propose a method of linear combination with dynamic weight for event duplication detection, combining the similarity score of arguments of time, location, subject and the similarity score of the remaining arguments dynamically to get the similarity score of event mention pairs. The experimental results show that the method further improves the results of the event duplication detection.

  • 出版日期2014

全文