Event Extraction in Vertical Domain Based on Similar Semantics and Dependency Syntax

Event extraction has become an unavoidable bottleneck in the process of event map construction and event analysis and prediction. The accuracy and completeness of the extracted determine the quality of event knowledge map construction and the credibility of event prediction. Aiming at the difficulties of extracting events based on traditional pattern matching, the low recall rate and accuracy rate, and the low accuracy of event extraction based on in-depth learning method in specific fields, this paper proposes a method for extracting news events in vertical domain based on similar semantics and dependency syntax, and the algorithm is verified by taking the field of political diplomacy as an example. By calculating the similarity between the trigger word set and the Syntax descriptor, the trigger word list is extended to lay a foundation for identifying event types accurately. Further, based on the guidance of pattern, through the construction of meta-event template, we can recognize and extract events in specific domain, so as to achieve a structured description of events. The accuracy of extraction results is obviously better than that of end-to-end event extraction model based on deep neural network, and it can be used for reference and implementation in other specific fields. Firstly, this paper makes theoretical and practical research on news sentence level meta-event extraction, and introduces in detail the methods of event extraction in political diplomacy based on similar semantics and dependency syntax. Finally, we summarized the main difficulties in event extraction, and discussed the application prospects of event extraction.