| With the rapid development of the Internet and e-commerce platforms,online shopping has become the most popular way of shopping at the moment.As a result,the number of comments on various e-commerce platforms and review sites has also increased.In this context,the explosive growth of information and comments in social media began to increasingly influence the decision-making behavior of consumers and producers.For a long time,traditional sentiment analysis methods only focus on identifying the overall sentiment tendency in consumer reviews,while ignoring important details in the reviews.They are only coarse-grained sentiment analysis and cannot specifically analyze consumer preferences and product pros and cons.This article takes the catering industry review text as an example,and analyzes the detailed information of the review text through the fine-grained sentiment analysis method,so as to obtain the attention tendency of user reviews and improve the service level of catering industry and the competitiveness of enterprises.First of all,this paper perform joint extraction and matching of keywords and emotional words.For the existing models,keywords and emotional words are often extracted separately,and the joint appearance of keywords and emotional words is ignored.Based on the existing model,this paper combines pre-trained word vectors to jointly extract the keywords and emotional words of the review text,and solves the problem that the original model cannot match the keywords and emotional words.By constructing a distance matching model,the two are matched,and the direct and indirect connection between emotional words and keywords are effectively obtained.Secondly,this paper reviews about the establishment standards and principles of the relevant service evaluation index system by consulting domestic and foreign documents.Based on the SERVQUAL model and fully understanding the service characteristics of the catering industry,the dimensions of the catering industry service evaluation system have been established.The indicators are food quality,restaurant equipment,restaurant environment,service level and price level.Through word frequency statistics and word2 vec,a five-dimensional keyword thesaurus was constructed,and the keywords were classified according to the thesaurus to obtain the attention tendency of the review text in the five dimensions.Finally,the text conducts an empirical analysis on the yelp review text,analyzes the frequency and content of keywords and emotional words in various dimensions,and finds that the service evaluation dimensions that consumers pay most attention to in the catering industry are food quality and service level.Customers generally pay more attention to specific content such as food taste and freshness of raw materials.The review text is analyzed based on the difference between positive reviews and negative reviews,and suggestions for improvement and related measures are put forward based on keywords and emotional words of various dimensions. |