Research And Application Of Social Recommendation Based On Field Trust And Distrust Influence

In the current era of the computer and network development,there is a large number of data with large dimensions every second.Faced with the problem of the information overload,people can use the recommender system to select useful information quickly and accurately.But in practice,there are problems such as the data sparsity and cold-start.With the explosion of the social network scale and user historical behavior data,the researches of the social recommender system has become popular in recent years.However,the existing social recommender algorithms rarely take into account the distrust relationship of users and the relevance of the social relationship to the field of user ratings.According to the domestic and foreign researches in the social recommendation,this thesis will integrate the distrust relationship in social networks and take into account the global influence and the field of the trust relationship.Use the distrust relationship to correct the social relationship.Moreover combine the concept of the group recommendation to cluster users.Based on the user's rating preference and user clustering results,the field correlation between users is calculated.The trusted friends of the target users is filtered by considering the global influence with the distrust relationship,and the experiments show that the proposed algorithm can improve recommendation quality and alleviate cold start problem.Finally,as the core module of the recommendation system,this algorithm will be used to build a social recommendation prototype system.This thesis mainly includes the following work:(1)Discuss the development background of the research work in this thesis.Introduce the research status of the domestic recommendation system firstly.Then summarize the current problems in the field and clarify the research focus of this paper and the significance of the research.On this basis,the emphasis and significance of the research are expounded.(2)In view of the neglect of the distrust relationship in traditional social recommendations and also considering the field relevance of the trust relationship and the influence of the distrust relationship on the user's global influence,users' social friends are filtered and the influence of users' social relationship is corrected.So on this basis,this thesis proposes a singular value decomposition algorithm based on field trust and distrust.(3)For the concept of the group recommendation,cluster the users based on the rating preference and improve the calculation method of the field relevance of trust according to the clustering result.Based on the results of the experiment,verify the effectiveness of the algorithm in improving the quality of recommendations.(4)Based on the above social recommendation researches based on the field trust relationship with users clustering and the distrust relationship,design and implement a recommendation prototype system.Describe the overall framework of the system and the main function modules in detail.