Dialogue text-oriented event extraction method and system

Abstract

The invention relates to a dialogue text-oriented event extraction method and system. The method comprises the steps of s1, periodically obtaining a dialogue text set; S2, filtering the dialogue text set twice to obtain an event-related dialogue text set; S3, creating an event template, and in the event-related dialogue text set, according to the event template, performing event category division through trigger words in the event template to obtain candidate events; and performing event element identification on the candidate events to realize event extraction from the dialogue text set. According to the dialogue text-oriented event extraction method and system provided by the invention, event extraction is carried out based on combination of pattern recognition and a machine learning method, so that the event template compiling cost is saved, the data sparsity is reduced, and the event extraction accuracy is improved.

Description

technical field [0001] The invention belongs to the field of text recognition and machine learning, and in particular relates to a dialogue text-oriented event extraction method and system. Background technique [0002] Event extraction refers to extracting event information from unstructured text, specifically extracting event components such as event trigger words, people, places, etc., and finally presenting it as structured text. As one of the important research directions in the field of information extraction, event extraction is widely used in public opinion analysis, information retrieval and other fields. [0003] Event extraction often adopts the following two methods: event extraction method based on pattern recognition and event extraction method based on machine learning. The method of event extraction based on pattern recognition uses pattern matching algorithm to match sentences and templates to realize event recognition and extraction. Pattern matching cons...

AI Technical Summary

Problems solved by technology

[0006] 1. The logical coherence of the context of the dialogue text is weak, and the expression of the dialogue text is more colloquial, the language is chaotic and complex, the length of the text is not fixed, and the event elements are scattered in different time slices, making event extraction more difficult;

[0007] 2. In the traditional event extraction method based on pattern recognition, the template definition is rigid, and most of the extracted objects are news texts. The adaptability and portability of event extraction for dialogue texts are poor;

[0008] 3. The traditional machine learning-based event extraction method has serious data sparseness and low accuracy

Images