Research On Recommendation Method Integrating User Trust And Similar Relations In Recommendation Socialized Commerce

In the era of big data,it is difficult for users to find the information they need from a large amount of data.Personalized recommendation algorithm can recommend the information for users,so it is widely concerned.Collaborative filtering recommendation is the most commonly used recommendation algorithm.Due to the lack of user rating data,collaborative filtering has the problems of sparse data,cold start and so on.On this basis,adding user social network information to the recommendation algorithm can alleviate the above problems.At the same time,the traditional recommendation system assumes that all users have the same importance,and there is no difference between them.However,it makes sense that some users are actually more important than others.On the other hand,with the development of e-commerce and social networks,a new e-commerce model,namely social commerce,has emerged.In the social business website,people can comment on the projects they are interested in.At the same time,users can establish social relationships and add users they trust to the "fruends" list.At the same time,in the social commerce environment,people are more likely to adopt the advice of a trusted friend.The opinions of the trusted friends will affect the purchase decision of the user.Therefore,in the social commerce environment,when recommending projects to users,the opinions of friends trusted by users should be considered.On the one hand,based on the common recommendation methods,the importance of users is not fully considered.At the same time,the accuracy of recommendation needs to be improved,and users need more reliable recommendation.On the other hand,the platform needs to discover the potential consumption information of users,provide users with better personalized products and services,and improve business profits.This paper proposes a recommendation method STPMF based on the user importance fusion of user trust and similarity.This method uses the user's rating data and social relationship data at the same time,measures the recommendation weight between users and selects the neighbor users through the user's similarity importance and trust importance fusion of similarity and trust between users.At the same time,considering the user,s own preferences and the influence of neighboring users,the project is evaluated,predicted and recommended.The main innovations of this paper are as follows:First of all,based on the weighted average algorithm based on trust,an improvement is proposed,which makes the local trust of users affected by the shortest path length between users.The longer the propagation path between users,the smaller the local trust between users.Secondly,considering the different importance of users in the recommendation system,the user importance is introduced to integrate the user similarity and trust.According to the similarity and trust importance of users,the proportion of similarity and trust among users in the user recommendation weight is balanced to improve the ability of identifying neighbor users.The main contents of this paper are as follows:First,select the user's neighbor user for recommendation.First of all,based on the social network graph model and the transmission property of trust,we model the trust relationship,calculate the local trust and the global trust between users,and integrate them to calculate the user trust.Secondly,the similarity between users is obtained by using user item rating data.Then,the recommendation weight is set,in which the similarity and trust between users are fused by the similarity importance and trust importance of users.Finally,the neighbor user is selected,and the user set w:ith the largest value is selected in the recommendation weight matrix.Secondly,the influence of user's own preference and neighbor's preference is integrated into the factorization framework of probability matrix,and the eigenvectors of unknown users and projects are updated by gradient descent method,and the optimal parameters and results of the algorithm are further obtained.Finally,the results of the proposed algorithm are compared with those of the classical algorithm in the real data set,and the conclusion is drawn that the proposed recommendation method based on user importance flusion of user trust and similarity can effectively identify neighbor users and improve the accuracy of recommendation.