A Novel Application Recommendation Method Combining Social Relationship And Trust Relationship

In the environment of mobile internet,mobile application information overload is an urgent problem to be solved.Recommendation system is still the key to solve the problem.Traditional collaborative filtering methods have the problems of data sparsity,cold start and so on.And the similarity of users is often calculated with Cosine and Pearson methods.When the nearest neighbor doesn't comment the predicted item,then the nearest neighbor has no influence on results,thus affecting the accuracy of collaborative filtering recommendation.In this paper,we have comprehensively consider social relationship,user preference and trust relationship,and put forward a novel application recommendation method that combines users' social relationship and trust relationship.Specifically,we combine social relationship,user clicks,tags and user preference towards applications to calculate similarity score,and fuse the trust relationship based on familiarity and user reputation to calculate trust score.The final prediction score is calculated by fusing similar relationship and trust relationship properly.And the proposed method can effectively improve accuracy of recommendations.The main research contents include the following aspects:(1)In this dissertation,we design a novel method to calculate user preference towards applications instead of using the frequency of usage directory.The conventional method does not consider some users use the applications with less cumulative time.In this case,we can't judge whether the user is actually fond of the application.In this thesis,we comprehensively consider the frequency of usage and the cumulative time,and with weighted sum method to calculate user preference.The proposed method alleviates the data sparsity.(2)In this dissertation,we propose a predicting model based on social connection among users.Our model comprehensively combine the user preference,the similarity in social network,social interaction among users and the features in WeChat.(3)In this dissertation,we propose a predicting model based on trust relationship among users.In order to get the trust value based on social network,we study and analyze the trust propagation mechanism.In order to get the trust value based on user reputation,we build the user profile.Finally,we combine these two aspects to get the user trust relationship prediction model.