Aspect-level sentiment analysis method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an aspect-level sentiment analysis method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining a target sentence to be subjected to aspect-level sentiment analysis and aspect words in the target sentence; based on the similar structure data set, performing aggregation feature processing on the target sentence and the aspect words to obtain fusion feature information corresponding to the target sentence; performing feature conversion on the fused feature information by using a bidirectional long-short-term memory network to obtain hidden state features corresponding to the target sentence; preprocessing the target sentence and the aspect words to obtain a dependency graph corresponding to the target sentence and position coding features corresponding to the words in the target sentence; and inputting the hidden state features, the dependency graph and the position coding features into a pre-trained graph convolutional neural network to obtain an emotion analysis result corresponding to the aspect words. According to the embodiment of the invention, the accuracy of the sentiment analysis result can be improved.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, in particular to an aspect-level sentiment analysis method, device, electronic equipment and storage medium. Background technique [0002] ABSA (Aspect Based Sentiment Analysis, aspect-level sentiment analysis) is a kind of task of sentiment analysis, which belongs to fine-grained sentiment analysis. It mainly identifies the emotional tendency of the subject of the comment on the aspect by analyzing the description of the specified aspect in the text. For example, if figure 1 As shown, for the sentence in the text: The food is good, but it certainly isn't a relaxing place to go, where the emotional word "good" indicates that the text has a positive (Positive) emotion for the entity "food" Tendency, while "isn't a relaxing place" implies that the emotional tendency towards the Atmosphere (environment) is negative. [0003] The existing method for aspect-level sentime...

AI Technical Summary

Problems solved by technology

[0004] However, the inventors found that the existing method of using the sentiment dictionary to judge the emotional tendency corresponding to the specified aspect of the subject has a strong dependence on the sentiment dictionary, and the sentiment dictionary is the result of artificial coarse-grained statistics, and the accuracy is not high. Make the result analysis of the emotional orientation corresponding to the specified aspect of the subject judged by the emotional dictionary inaccurate

Images