Research On Trust Recommendation Model Based On Sentiment Analysis For Comment Text

With the rapid development of information technology and the Internet,e-commerce has become an indispensable part of people’s daily life. However, while e-commerce systems offer more and more choices for users, the flood of the commodity information means that users need to pay more time and energy to find the goods they need. In a certain degree, the excess information means the lack of information, the e-commerce recommendation system came into being under such background. E-commerce recommendation system through the simulation of sales staff to recommend products to users, the user will be converted to buyers, and can improve the user’s satisfaction and loyalty. Due to the huge potential commercial values,e-commerce recommendation system has been studied extensively.Existing recommendation systems usually focus on the project evaluation data and user’s historical behavior information, while ignoring the rich information contained in the product’s review data. And most of the recommendations given by the existing recommendation systems are based on the whole of the goods, and neglect the different requirements of the user’s for the various aspects of the commodity. To solve these problems, this paper presents a credible recommendation method based on sentiment analysis of the comment texts. First of all, this method through feature extraction and sentiment analysis of the reviews to obtain the emotional tendency of different dimensions of the product, so that the user can have an intuitive understanding of the characteristics of the commodity. On this basis, the D-S evidence theory is used to analysis the results of the emotional analysis based on the fusion of the feature categories, and obtain the credit score of different evaluation dimensions. Finally, according to the total score of credit, as well as the user’s credit score of different evaluation dimensions, to make recommendations for the user.The credit score obtained by D-S evidence fusion, which includes all the characteristics of the user’s comments on the results of the analysis of the results, is a comprehensive, reasonable and credible score. Relying on this score to make recommendations can get higher user trust, and can make a reasonable explanation forthe recommended results. Moreover, according to the different characteristics of the user’s attention, we can make personalized recommendation.