Text classification and device

Abstract

The invention discloses a text classification and device, pertaining to the technical field of computers. The method comprises the following steps: determining word vectors corresponding to keywords according to a word vector mode as for each keyword of many keywords incorporated into a keyword library of a business information library; determining potential expansion words of keywords on the basis of the word vectors corresponding to keywords; adding the potential expansion words into the keyword library as well as adding an expansion rule to a matching rule library when receiving the inputted expansion rule for the potential expansion words and detecting addition instructions of the potential expansion words; determining first probability of each pre-set class in multiple pre-set classes comprising to-be-classified texts according to a mode distribution classifier on the basis of the keyword library and the matching rule library; and determining a class to which to-be-classified texts belong from multiple pre-set classes on the basis of the first probability. The text classification and device can reduce labor cost of building a business information library and helps coverage rate and accuracy of text classification.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a text classification method and device. Background technique [0002] With the rapid development of computer technology, massive information resources exist in the form of text. Because massive information resources often contain a lot of real and valuable information, for example, when the information resource is the dialogue data between users and customer service in a certain business, the dialogue data often reflects the business development situation, business problem feedback, etc. Therefore, in order to quickly and effectively discover valuable information from massive information resources, it is necessary to classify the text corresponding to the information resources. [0003] At present, a text classification method is provided, specifically: technicians manually construct the keyword database and matching rule database of the business information base. After the c...

AI Technical Summary

Problems solved by technology

[0004] Since the keyword library and the matching rule library of the business information base are completely manually constructed, the construction time of the keyword library and the matching rule library is relatively long, and the construction efficiency is low, and the establishment and maintenance of the keyword library and the The labor cost of matching the rule base is relatively large

In addition, since the keyword library and the matching rule library are completely dependent on the experience of technicians, the accuracy and coverage of the above text classification methods are relatively low

Images