Research On Sentiment Analysis Method Of Comment Text

Sentiment analysis is an important task in the field of natural language processing.It aims to extract the emotional content of the text.Depending on the granularity of the analysis,it can be divided into the analysis of the article,the analysis of the sentence,and the analysis of the aspects in the text.Sentiment analysis on all aspects of the text is finegrained sentiment analysis.At present,the technology of analyzing the emotion of the whole article is relatively mature.It has achieved good results.However,only analyzing the overall emotional tendency of the text will cover up many details in the text.The overall emotions of the text cannot express people’s emotional tendencies towards different aspects.Only paying attention to the overall sentiment of the text and ignoring specific details may affect the results of the analysis.For comment texts,the same text may have different emotional tendencies in different aspects.Sometimes it is necessary to consider the sentiment of the comment text in a certain aspect.For example,when a user purchases a product,it is necessary to focus on analyzing the emotional tendency of the product in terms of quality if the quality of the product is what he values most.This paper mainly studies the sentiment analysis of comment text,and realizes the fine-grained sentiment analysis of comment texts.The fine-grained sentiment analysis of comment text has problems such as high computational complexity and poor prediction accuracy.In response to these problems,this paper proposes a sentiment analysis model that combines character vector representation and attention mechanism for the sentiment analysis of comment texts.The main work of this paper is as follows:First,a word vector representation that combines Word2 vec and Fast Text is proposed to better obtain the information of the input text,using a neural network based on GRU,CNN,and an Attention mechanism of maximum pooling and average pooling.The model extracts deep-level text features and realizes the fine-grained sentiment analysis of comment text.This paper improves the neural network model of sentiment analysis,and realizes the task of fine-grained sentiment analysis of comment text.The improved sentiment analysis model is applied to the AI Challenger 2018 fine-grained user comment sentiment analysis data set.After comparison and analysis with commonly used sentiment analysis models,the model proposed in this article has achieved good analysis results.Secondly,based on the improvement of the algorithm,the sentiment analysis system of the review text has been realized in this paper.The system can judge the emotional tendency of the review text in some aspects,filter the review text according to different aspects,and display it intuitively,so that the user can obtain the emotional information of the text more quickly.