A tag-based document classification method, system, device and storage medium

Abstract

A label-based document classification method, system, device and storage medium provided by the present invention, the method includes the following steps: obtaining the text to be classified and the label, performing word segmentation on the text to be classified to obtain a word embedding vector; Segment the word to get the tag embedding vector; merge the word embedding vector to determine the first embedding sequence; get the sentence embedding vector according to the tag embedding vector and the word weight coefficient in the first embedding sequence; merge the sentence embedding vector to determine the second embedding vector, according to The label embedding vector and the sentence weight coefficient in the second embedding sequence are used to obtain the text embedding vector; the classification probability of the text to be classified is determined according to the text embedding vector, and classification is performed according to the classification probability; the document representation vector adopted by the method is more distinguishable, so that Obtain more excellent classification performance, and can be widely used in the field of natural language processing technology.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, in particular to a tag-based document classification method, system and storage medium. Background technique [0002] With the development of computing technology and the popularization of the Internet, network information resources are growing explosively. In the face of massive and chaotic network information, the problem faced by users is no longer how to obtain information, but how to obtain large-scale information resources. In order to efficiently and accurately find information that can meet their own needs, realize intelligent and efficient classification of massive network information, accurate identification of user interests, and personalized information recommendation. Therefore, the labeling and classification of information is of great significance. Existing text classification methods include the following categories: [0003] 1. The method of manual la...

AI Technical Summary

Problems solved by technology

[0003] 1. The method of manual labeling, which works well in a small amount of data and low latitude, but the labeling efficiency is low, the cost is high, and the label dimension is too simple, which is difficult to meet the explosive growth of text classification requirements in the Internet field;

[0004] 2. Traditional machine learning methods, such as naive Bayesian, SVM, etc., have good results, but rely on handcrafted rules or features, the workload is heavy, and information omission and redundancy are prone to occur, which affects the classification effect;

[0005] 3. The existing deep learning methods, such as fasttext, textcnn, etc., can automatically learn relevant useful features, but they often only focus on using the content information of the text to extract features, while ignoring the inherent writing structure of the document

Images