Automatic extraction method for case semantic elements

Abstract

The invention discloses an automatic extraction method for case semantic elements. The automatic extraction method comprises the following steps: (1) carrying out part-of-speech tagging on a brief case text by virtue of a sequence tagging module according to a pre-trained Chinese sequence tagging model, so as to recognize nouns, verbs and gerunds in the brief case text; (2) enriching a seed lexicon through word meaning similarity algorithm by virtue of a word meaning similarity module; (3) carrying out manual correction for several times by virtue of an adjusting-optimizing module, so as to further adjust and optimize an element extraction result; and (4) creating a classifier library through the elements according to the category by virtue of a classifying lexicon module. According to theautomatic extraction method, the case semantic elements are automatically extracted from a detailed, complicated and ruleless brief case, the semantic element extraction efficiency of the case is greatly improved, and the manual extraction cost is lowered; and meanwhile, a system can be applied to semantic comprehension analysis of other unstructured text information of public security industry,and an effective method is provided for correlation analysis of the cases, so that the case treating efficiency is improved.

Description

technical field [0001] The invention relates to the technical fields of natural language processing and machine learning, in particular to an automatic extraction method of case semantic elements. Background technique [0002] With the rapid development of natural language processing and machine learning, these technologies have been widely used in search engines, intelligent translation, machine question answering, user recommendation, public opinion monitoring and other related industries. Public security organs have accumulated a large amount of data information in the long-term informatization process, but the application of data at this stage is still limited to simple query of structured data, and has not yet carried out in-depth application of unstructured or semi-structured data. At the same time, there are still many problems in these unstructured or semi-structured data: the information format is not uniform, and the information description is diversified. Based o...

AI Technical Summary

Problems solved by technology

[0007] Aiming at the deficiencies of the prior art, the present invention provides an automatic extraction method for the semantic elements of a case, which solves the problem of dimensionality disaster, especially when used as the input of a deep learning algorithm, the performance of the algorithm is seriously affected; there is a "lexical gap" Phenomenon, that is to say, the problem of not being able to describe the similarity between words well

Images