| Social e-commerce refers to a new type of e-commerce in which users spontaneously share and disseminate information through content generation,interactive interaction,etc.,so as to realize the marketing promotion and sales behavior of products or services.The massive amount of generated content on the platform makes it difficult for users to find the information or products they really need,and malicious recommendations,false publicity,phishing and other dishonest behaviors on the platform make users doubt the authenticity of the information evaluated by others.These problems make the platform face double loss of profit and reputation.Therefore,it is very important to establish a recommendation mechanism for social commerce to help alleviate the problems of information overload and lack of trust on the platform.This paper focuses on the above problems,carefully sorts out and summarizes the research results of domestic and foreign scholars on the establishment process and influencing factors of social business recommendation mechanism,and finds that the recommendation algorithm fails to fully consider the explicit and implicit trust between users and does not describe the individual interests of users deeply enough.First of all,from the perspective of implicit trust,this paper extracts and quantifies the three factors that reflect the potential trust relationship of users to build a user trust fusion model to effectively identify the set of trusted neighbors of the target user,and introduces the product popularity as an adjustment factor to predict the ratings of target users.The experimental results show that the model can effectively achieve better recommendation effect by carefully describing the multi-dimensional implicit trust between users in the recommendation process,and the recommendation performance can be further improved by introducing the popularity factor of the product itself;Secondly,Explicit trust directly reflects the attitude or preference of users,and plays a very important role in trusted neighbor screening.The direct trust and distrust evaluation data between users is analyzed to construct the network topology,and the trust transfer characteristics in the network are combined to measure the explicit trust between users.Construct a user explicit and implicit trust fusion model,and use the integrated overall trust value as the basis for neighbor user selection to implement a collaborative filtering algorithm based on explicit and implicit trust and product popularity.The experimental results show that the malicious nodes in the network can be effectively identified by introducing the distrust evaluation into the social network,thereby improving the objectivity and accuracy of the trust measurement results,and the proposed fusion recommendation mechanism can effectively improve the ability to identify trusted neighbors by comprehensively considering the explicit and implicit trust between users,thereby improving the accuracy of the recommendation results;Finally,considering that the user’s acceptance of the recommended products or services is affected by their own interests and preferences,based on the above application of trust recommendation,this paper introduces the time weight function into the product category label data,and applies the time window method to quantify The user’s own dynamic interest in different products,and the final weighted balance between the user’s personal interest preference and trusted friends’ preference produces the recommendation list.The experimental results show that the recommendation mechanism weighted by a certain proportion of users’ personal interests score and social trust score is better than the three compared recommendation models in terms of accuracy,recall and F value,which helps to improve the shortcomings and drawbacks of a single recommendation scheme,thereby improving the recommendation effect and providing certain theoretical guidance for the platform to carry out personalized recommendation business. |
