Multi-label text classification method based on mixed attention mechanism

Abstract

The invention discloses a multi-label text classification method based on a mixed attention mechanism. The method comprises the following steps: S1, constructing a multi-label classification model based on the mixed attention mechanism; s2, inputting a to-be-classified text into the multi-label classification model; s3, in the multi-label classification model, sequentially carrying out word embedding and coding processing on the input text, extracting text feature representations related to the content of the text and text feature representations related to labels corresponding to the input text in parallel, fusing the text feature representations and mining label relations; s4, obtaining a multi-label text classification result based on the label mining relation mining result and the fusion text feature representation. According to the method, the text feature representation for each label can be obtained; meanwhile, the features of the text are extracted by using a self-attention mechanism, each word in a text sequence can establish a relation with words at any distance in the sequence, and the problem that CNN and RNN depend on the modeling ability for a long distance is solved.

Description

technical field [0001] The invention belongs to the technical field of text classification, and in particular relates to a multi-label text classification method based on a mixed attention mechanism. Background technique [0002] With the popularization of mobile devices and the rapid development of information technology, the Internet has developed rapidly in an unprecedented manner. Internet applications represented by Weibo, Taobao, WeChat, and Zhihu, etc., generate massive amounts of data every day. Human beings have entered the era of big data. Among them, text is an important information recording method in the human world. Nowadays, various text forms such as emails, chat records, and comments exist on the Internet. Most of these texts are unstructured texts with the characteristics of messy content and complex structure. The traditional text classification method of establishing rules in a timely manner has been unable to process these information efficiently. The ...

AI Technical Summary

Problems solved by technology

[0005] In view of the above-mentioned deficiencies in the prior art, the multi-label text classification method based on deep learning provided by the present invention solves the problem of unsatisfactory classification effect in the existing text classification method

Images