Text classification method and device and storage medium

Abstract

The embodiment of the invention discloses a text classification method and device and a storage medium. According to the embodiment of the invention, the method comprises: obtaining to-be-classified text pairs; wherein the to-be-classified text pair comprises a first text and a second text; selecting a first keyword of the first text and a second keyword of the second text from the to-be-classified text pair; performing feature extraction on texts in the to-be-classified text pair; then, according to the word feature information of the first text and the word feature information of the secondkeyword, obtaining a second text; obtaining first attention feature information, and obtaining second attention feature information according to the word feature information of the second text and theword feature information of the second keyword, and then classifying the to-be-classified text pairs based on the first attention feature information, the second attention feature information, the first semantic feature information and the second semantic feature information to obtain a classification result. According to the scheme, the text classification accuracy can be effectively improved.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a text classification method, device and storage medium. Background technique [0002] With the development of artificial intelligence (AI), natural language processing technology has become an important research direction in Internet applications. In the Natural Language Processing (NLP) method based on deep neural network, the classification of text pairs is a very basic and important problem, which is widely used in search, recommendation, dialogue and other fields. Usually, in a pair of texts, text A refers to user questions, and text B refers to content sources to be matched, such as questions in the question-and-answer library, content of web pages, text descriptions of products, and so on. Text pair classification should not only consider the information of a single text, but also consider the interaction information between text pairs. At present, many methods si...

AI Technical Summary

Problems solved by technology

At present, many methods simply combine the features of the text on both sides, or directly input them into a model with a self-attention mechanism for classification, and the classification effect is poor.

Images