Text information classification method and device based on combination learning, and computer equipment

Abstract

The invention relates to a text information classification method and device based on combination learning, and computer equipment. The method comprises the steps of S1, inputting first text information into a bi-directional long-short-term memory recurrent neural network Bi-LSTM for performing processing; S2, by taking an output after processing by the bi-directional long-short-term memory recurrent neural network Bi-LSTM as inputs of a long-short-term memory network LSTM and a collaborative convolutional neural network CNN, jointly performing mixed neural network training to obtain a classification network of text information; S3, inputting second text information into the bi-directional long-short-term memory recurrent neural network Bi-LSTM for performing processing, and inputting a processing result into the classification network, thereby obtaining classification of relations between information entities of the second text information. The problem of difficult classification / sorting of existing text information due to numerous types and large data amount is solved. The information entities such as "personnel", "place", "time" and the like in the text information can be automatically recognized, so that the time cost of the case handling personnel is remarkably reduced.

Description

Technical field [0001] The invention relates to the technical field of big data analysis, in particular to a method, device and computer equipment for text information classification based on joint learning. Background technique [0002] As a basis for the research and judgment of criminal acts, textual intelligence plays an important role in case detection. Especially in the context of big data, the acquisition of textual intelligence such as website social data, communication content, and chat records greatly enriches the research and judgment information. Case detection provides solid information support. However, with the rapid increase of textual intelligence, the problems of difficult information analysis and utilization have become increasingly prominent. "Large amount of data, low value density, and low utilization" have become difficult points for textual intelligence analysis. [0003] Text intelligence extraction has become an important way to solve the above problems. ...

AI Technical Summary

Problems solved by technology

[0006] The present invention aims at the problem that the existing text information is difficult to classify / comb due to the many types and large amount of data, and provides a text information classification method, device and computer equipment based on joint learning, which can automatically identify the " Information entities such as "person", "place" and "time", from which the relationship between people and people, people and places, and people and events can be extracted, while making full use of the diverse text intelligence under the background of big data, it can significantly reduce the time cost of investigators

Images