Research On Dynamic Fuzzy Trust Relationship And Its Application In Social Commerce Recommendation

Social e-commerce based on social media has become a new trend in the development of ecommerce.Social commerce is prone to high-risk perception due to the inequities of online and offline,which makes it difficult for users to select from massive information.The traditional recommendation algorithm faces the problems of data sparsity and cold start,and it is difficult to further improve the recommendation effect.Therefore,we can consider using social data to integrate social networks,especially trust networks into personalized recommendation,so that the social business recommendation based on trust can effectively improve the recommendation effect and the efficiency of e-commerce activities.In this thesis,an indirect trust calculation method based on dynamic trust propagation and trust aggregation considering the time factor is proposed.Combining the dynamic trust relationship with the recommendation system,a social business recommendation model based on the dynamic trust is proposed.Finally,the validity of the model is verified by the actual data of Douban website.First,in order to solve the problem of trust propagation considering the time factor,the thesis introduces the historical fuzzy trust data considering the time stamp.The time decay function and the discount function are applied to deal with the time attribute of trust.We design a new propagation operator which is more consistent with the trust decay speed and satisfies the trust boundary preservation.The new trust propagation operator has better computational consistency in the propagation of long trust chain.Then,aiming at the problem of trust aggregation considering the time factor,this thesis summarizes the main classification and application scenarios of classical aggregation operators and combs the common aggregation methods extended to intuitionistic fuzzy set in the field of disjunctive and average class aggregation operators.By selecting appropriate aggregation operators for intuitionistic fuzzy set and combining the characteristics of trust aggregation path,this thesis proposes four single factor trust aggregation operators based IFWA considering four different factors: path length,path strength,information reliability and time decay and two-factor and multi-factor trust aggregation operators based IFHWA and IFOWOW through combing above four factors.Next,in view of the combination of dynamic trust and recommendation system,this thesis defines different calculation methods of implicit trust and user similarity in the context of social business,analyzes the similarities and differences of trust and similarity,and constructs a better user relationship by reconciling trust and similarity.Combined with the new dynamic indirect trust calculation process considering time factor and collaborative filtering recommendation,this thesis combs out the whole process model of social business recommendation based on dynamic trust.Finally,aiming at the application problem of social business recommendation model based on dynamic trust,this thesis combines the actual data of Douban website to apply.The thesis introduces the social business characteristics of Douban platform,selects data indicators according to the needs of the model,compiles crawlers to grab data,processes data and makes preliminary statistical analysis.Matlab is used to program and calculate the social business recommendation model based on dynamic trust,and the final recommendation results are obtained.Compared with the traditional collaborative filtering recommendation and the social business recommendation results based on static trust,the operability,effectiveness and advantages of the model are confirmed.The suggestions are provided for Douban platform to accelerate the commercialization process with the help of social business recommendation.