Book Recommendation Based On The Analysis Of User Comment Sentiment Orientation Algorithm Research

Nowadays,the Internet is not only a platform for disseminating information,but also an important channel for people to share information and express their opinions.Especially after the wide application of Web2.0 and mobile social networks,a large number of users are accustomed to leaving comments on various items and events on the Internet,thus all major websites have accumulated a large number of comment data sets.Such a large amount of comment information has caused confusion to users' decision-making and "information overload".Therefore,personalized recommendation technology has become a research hotspot among scholars.However,most of the current recommendation technologies only consider user rating data,personal basic information,etc.,but fail to make full use of the large amount of unstructured comment data with high value that can reflect the user's personal preference,and these comment data can alleviate the problem of data sparseness and low recommendation accuracy to a certain extent.Based on this,this paper delves into the content of user comments and proposes a personalized recommendation model based on the analysis of user comments based on the research status of emotion analysis technology and recommendation of reading social networking sites.The model is mainly divided into two modules:comment sentiment analysis and personalized recommendation.In the process of emotion analysis,the main syntactic dependency is firstly obtained through syntactic analysis.The feature word-view word pair is extracted based on the syntactic relationship.In the extraction process,the seed lexicon and synonym word forest of feature words are added for feature selection to improve the accuracy of feature extraction.Then,using the Chinese semantic similarity calculation method based on the HowNet dictionary,attribute feature clustering is carried out for characteristic words;Finally,considering the semantic differences caused by different combinations of degree adverbs and negative words,the modification percentage of degree adverbs is set,and a calculation method of emotional intensity is designed to obtain the emotional score of user attributes.In the personalized recommendation module,the user interest preference model is constructed by using the user's attention to the attribute characteristics and the public emotion rating.The similarity of the user attribute characteristics is calculated based on the user preference,and the user interest degree is obtained for recommendation.Finally,the application of the review data of "Douban Reading" was studied,and the optimization recommendation strategy for Douban reading platform was obtained.The effectiveness of the model was verified by detailed experimental design.