Research On Adaptive Recommendation Method Based On New Trust Model

With the rapid development of Web2.0 technology,more and more available information resources are available on the Internet,and it becomes more and more difficult to get the information they need from the massive data information.In order to solve the problem of information overload,the recommendation system arises at the historic moment,as a kind of information filtering technology,it is also a favorable technology to solve the problem of network information overload.However,with the application of the system continues to expand the scale,the number of users and projects increased dramatically,but the direct relationship between the user and the item proportion is more and more small,which makes the recommendation system faces severe sparsity problem,resulting in decreased quality of recommendation.Moreover,traditional collaborative filtering recommendation algorithm is the main user rating similarity as the only member information,ignoring the social relationships between users,some malicious users may provide a lot of false information will cause the score,the recommended results may deviate from the needs of users.Therefore,facing the problem of data sparsity and user profile injection attack problems,recommend how to improve the quality of recommendation system has become the focus of attention in the field of topic recommendation technology.Based on the comprehensive analysis of the current research situation at home and abroad,this paper makes a further research on collaborative filtering technology based on the data source credibility and neighbor selection strategy.First of all,in view of the shortcomings of traditional user similarity computing methods,an improved method based on project clustering is proposed to calculate user similarity.Based on the direct connection path in the trust network,the direct trust between users is obtained.Based on the non direct access path in the trust network,the indirect trust between users is calculated,and the expert recommendation based on the PageRank algorithm is calculated.The trust degree between users is calculated in a multi angle form,and a new type of improved user similarity is constructed.Trust computing model improves the accuracy of trust between users.Secondly,the calculation results based on user trust degree,this paper proposes an adaptive recommendation method,this method first determines the target user group,and then based on the number of target users where the group of neighbors,adopt different adaptive recommendation strategy;target user according to different classifications,take the similarity and different combinations of selected recent credibility the neighbor set of users makes in improving the neighbor users choose the right at the same time,the search scope to further expand the crazy neighbor users,so as to effectively overcome the problem of data sparsity.Finally,on the selected data sets,the proposed adaptive algorithm and the existing classic algorithms are compared and analyzed.