Diversity Recommendation Research Based On User Trust Relationship In Social E-commerce

With the continuous updating and iteration of science and technology,mobile Internet technology,communication technology and big data technology are gradually mature.With the rapid development of social networks,social relations are gradually becoming apparent.Due to the characteristics of mobility,immediacy and convenience,especially the consideration of user personalization,social commerce(SC)has become the main development product after e-commerce.Personalized recommendation system has a significant effect on strengthening user loyalty,improving user shopping demand and enhancing the core competitiveness of enterprises.However,due to the explosive growth of the amount of information in the network,social e-commerce shopping platforms often fail to provide users with a high-quality recommendation result due to the problem of “information overload”,resulting in the decline of user experience.Therefore,from the perspective of users’ social network,it is of great theoretical and practical significance to analyze the influence of social relationship on users’ consumption process in social e-commerce,so as to build a reasonable recommendation model for social e-commerce,which can enhance users’ stickiness,enhance users’ continuous shopping freshness and improve enterprise market share.Based on the latest literature of social network,product recommendation and social e-commerce,this paper discusses the difference and relationship between trust and reputation in the environment of social e-commerce.Based on the in-depth analysis of the influence of trust and reputation on users’ purchase behavior in social e-commerce,this paper proposes a relationship model between users and social e-commerce recommendation model,and reveals the mechanism and influence degree of users’ social relationship on the accuracy of social e-commerce recommendation model.Based on the above research,the overall framework of social e-commerce recommendation model based on user experience,which is composed of user trust and diversity,is designed to improve user experience.According to the meaning and characteristics of user trust relationship in SC,this paper constructs a social e-commerce recommendation model based on trust.The main contents include: using social reputation to deeply describe the role of user relationship in the recommendation system,ranking user reputation by using the degree of trust of users in social network,integrating user trust and reputation quantitatively by graph neural network,and combining the combined results The new matrix is constantly corrected to obtain more accurate user trust,and then the new scoring model is updated after matrix decomposition,and finally a more accurate prediction scoring matrix is obtained.Comprehensive current hot Considering its compatibility with the background of social e-commerce,this paper proposes a unique SC diversified recommendation model,which includes the following aspects:making full use of the long tail effect in the recommendation system to build a diversified recommendation list;reordering the recommendation list based on the score value and popularity of the project,and sorting the product score,popularity and novelty Three dimensions further optimize the recommendation list given to users;improve the diversity of recommendation without losing the accuracy of recommendation.Different from the traditional e-commerce recommendation method,the social e-commerce based recommendation method takes the trust relationship of consumers as an important consideration.The accurate measurement of trust relationship and users’ social reputation is very important for the accuracy of this kind of recommendation system.Therefore,there is one sidedness in the research only from the perspective of similarity or trust.Aiming at the defect that the current recommendation algorithm does not consider the measurement of user trust and social reputation accurately,this paper analyzes the reputation of user trust,adds diversity index to complete the recommendation results,proposes a personalized recommendation algorithm integrating user trust and diversity,expounds the relationship between trust and diversity,and uses extreme learning machine to synthesize user trust And diversity to get the final score of users.This method not only enhances the requirements of user preferences,but also obtains more stable trust measurement,and improves the integrity of recommendation results.The research on trust and diversity recommendation based on social e-commerce in this paper not only helps to meet the real-time personalized needs of users in social e-commerce,enhance the sense of user experience,but also helps to improve user stickiness,optimize the business level of e-commerce service platform,and enhance the market share of enterprises,which has vital reference value and practical significance to promote the rapid and good development of social e-commerce.