Sentiment Analysis Of Meituan Review Data Based On Fuzzy Comprehensive Evaluation Model

The shop's rating on the platform is a comprehensive score,which cannot explain the situation of the product from multiple perspectives.It is difficult for customers to choose a shop that meets their needs based on the comprehensive score.Existing evaluation models rely on experienced managers to select product attributes and weighted attribute values,which is subjective.In view of this,this paper mainly proposes a sentiment analysis model based on fuzzy comprehensive evaluation for the diversity of evaluation angles and the objectivity of the review results.The main work of this article is as follows:The rating of the store on the Meituan platform is a comprehensive score,and it is impossible to explain the situation of the store from multiple angles.It is difficult for customers to select a store that meets their needs according to the comprehensive score.The existing evaluation model is given by the experienced manager to give the choice of commodity attributes and the weight of the attributes,which is very subjective.In view of this,this paper proposes a sentiment analysis model based on fuzzy comprehensive evaluation for the diversity of evaluation angles and the objectivity of review results.The main work of this paper is as follows:(1)A model for analyzing emotions from multiple angles is proposed.The model adopts the method of combining keyword extraction and fuzzy matrix,extracts the keywords from the corpus of comments,and then sets the fuzzy matrix according to the weight of the keywords.The results of the evaluation of sentiment analysis are placed in the fuzzy matrix to construct a fuzzy comprehensive emotion evaluation model based on fuzzy matrix.The model gives the scores of other keywords in addition to the comprehensive score of the store,comprehensively analyzes the scores of the store's taste,environment,price and other factors,providing customers with a multi-angle selection guide.The resulting scores also provide good guidance to merchants to help them adjust their business models and strategies.(2)The sentiment analysis of Meituan platform reviews is mainly focused on the classification of emotions.Here are fewer discussions about different customer groups and different factors' emotions.For massive review data,it is also difficult to extract the key factors of reviews.The model can determine the consumer's attitude to one orseveral factors by extracting the keywords of the evaluation sentence and combining the fuzzy matrix method.This paper first analyzes the review data in plain text,the comprehensive evaluation results obtained by the method on the review dataset are only 4% different from the comprehensive score of the Meituan platform,which verifies the reliability of the model.The method also gives the sentiment scores of different keywords in the review data,and evaluates the product from multiple angles.(3)Emotional tendency analysis includes plain text and comment data with emoji.In view of the extensiveness and ambiguity of the current network symbols and expressions,this article extracts comments containing emoticons.First find the comment data that contains emoji then manually label the meaning of the emoticons,Finally,according to the emoticon dictionary,we replace the emoticon data in the comment with the corresponding emotive words,and then sentiment analysis was performed.Through experiments on massive comment data containing emojis,the optimal value of emotional tendency calculated based on the fuzzy comprehensive evaluation is taken as the final fuzzy evaluation result,and it is only 1% different from the average value of the user evaluation.Obviously,the comment data containing emoticons is closer to the user's true evaluation than the plain text comment data.(4)Different users focus on different product attributes,the model adds a personalized user review recommendation mode.When the user selects the keywords of interest in the system,the system will set the weight of the keywords based on the keywords selected by the user,and the comprehensive evaluation result will be calculated from the weight and fuzzy matrix.The user can refer to this result to make consumption select.