Research On Key Comment Extraction Methods Combining Sentiment Analysis And User Behavior

After Web 2.0,people have become accustomed to posting their opinions on social networking sites,such as reviews of attractions,reviews of specialties,hotel reviews,or The experience of a branded product.Therefore,user network comments are open,available,and shared.They are an important form of public opinion.Users are also accustomed to retrieving important information from the review texts of products and finding an accurate reference for their own decisions.However,at present,the amount of information on the Internet is large.For example,in the e-commerce system,there are thousands of evaluation contents for a certain product.How to find a viewpoint that meets their needs for these comments is particularly important but difficult to implement.Therefore,how to accurately extract key information is very practical.At present,there are two problems in the key information extraction methods:(1)At the present stage,when calculating the emotional polarity of sentences,most of them consider the emotional polarity of pure text sentences,and rarely consider the emotional polarity of the expressions.The influence of emotional polarity makes the calculation of emotional polarity of sentences inaccurate;(2)Most scholars analyze only the surface behaviors such as number of comments,likes and number of forwards when analyzing user comments,and rarely comment on the content.As well as responding to the content of the comments for consideration,it is clear that the one-sidedness of the analysis has a great bias in the extraction of key information,making the calculation results unsatisfactory.Therefore,in order to solve the above problems,this paper proposes a key comment extraction method that combines sentiment analysis and user behavior.The main work is divided into the following two parts:(1)An emotional key sentence extraction method based on emoticon analysis is proposed.This method improves the method of emotional key sentence extraction.When calculating the sentence emotional polarity score,it focuses on the influence of the emotional polarity of the expression on the sentiment polarity of the sentence,making the calculated sentence emotional polarity more accurate;(2)The method of extracting commentary key sentences that incorporates comment content and response support rate is proposed,which solves the lack of consideration of the quality of product review content and the attitude of users who reply to the comment content,and considers the quality of the comment content.It also considers the attitude of supporting the comments in the reply comments,calculating the support rate and the praise rate of the reply comments,thereby improving the efficiency of extracting the key sentences of the comments.Emotional key sentence extraction method based on emoticon analysis,extraction method of comment key sentence with fusion comment content and reply support rate,and key comment extraction method of fusion sentiment analysis and user behavior obtained by combining the two methods are effective.Experimental verification.The experimental results show that the method of extracting emotional key sentences based on emoticon analysis has higher accuracy,recall rate and F value than other mainstream methods,which can extract emotional key sentences more accurately.The method of extracting the key sentences of the reply support rate can be more suitable for the user's real emotions;the combined method of extractive sentiment analysis and user behavior is more comprehensive and the extraction effect is better.