Text sentiment analysis method based on BERT model and double-channel attention

Abstract

The invention discloses a text sentiment analysis method based on a BERT model and dual-channel attention, and the method comprises the steps: building a sentiment language library, carrying out the word segmentation of text data, extracting sentiment information words in each piece of text data, and providing dual-channel input containing semantics and sentiment information words for model training; extracting sentiment information words contained in the text statements to obtain a sentiment information set corresponding to each text statement; constructing a word vector double-input matrix of semantic information and sentiment information by utilizing a BERT model; a dual-channel attention feature extraction module composed of a BiGRU neural network and a full-connection network is addedto a hidden layer, and the emotion semantic capturing capacity of the model is enhanced; and fusing the obtained deep semantic information with the emotion information word vector to obtain a final deep semantic expression. According to the method, the statement-level text sentiment analysis performance is effectively improved, the superiority of the method is verified through experimental comparison, and the method has good performance in multiple evaluation indexes.

Description

technical field [0001] The invention relates to the technical field of natural language processing, in particular to a text sentiment analysis method based on a BERT model and dual-channel attention. Background technique [0002] Text sentiment analysis at the sentence level in the text, that is, sentiment analysis for sentences, is the process of analyzing, processing, inducing and inferring subjective texts with emotional color. With the development of social media such as forum discussions, blogs and Twitter, there are massive emotional data, so sentiment analysis technology is playing an increasingly important role. The deep learning model based on neural network can learn the distributed vector representation of words. This low-dimensional and continuous word representation overcomes the shortcomings of traditional word representation methods and can be used as a good example of other deep neural network models. Input, through the continuous learning of multi-layer net...

AI Technical Summary

Problems solved by technology

Despite the usefulness of emotional language words, the application of emotional language knowledge has been limited in recent years in deep neural network models such as convolutional neural networks (CNN) and long short-term memory networks (LSTM)

Images