Research And Design Of Collaborative Filtering Recommendation Based On Optimal Trust Path

The developing information technology not only provides people with tremendous internet resources,but also leads them into “information overload” situation.How to help users find valuable information in a high speed is the core problem of recommendation technology.Collaborative filtering recommendation algorithm is the most extensive among recommendation technologies.However,with the advancement of online social network and the complication of social network,the trust between users affects recommendation results.The traditional collaborative filtering recommendation algorithm has the following problems:(1)The sparsity of data.The sparsity of user rating data results in that the degree of similarity calculation and the rating prediction is not accurate and users can't get the suitable recommendation information.(2)Vulnerability.The open recommendation system allows users to rate or comment freely.However,some users may offer fake information which lead to the unreliable recommendation results and can't provide other users satisfied recommendation.(3)Do not consider trust relationship.The traditional algorithm only cares users' rating data and don't consider the trust among different users and the worth of the trust.Considering that the problems and challenges within collaborative filtering recommendation algorithm,this paper proposes an improved collaborative filtering recommendation algorithm,which mainly studies:(1)Considering the data sparsity of collaborative filtering recommendation algorithm,taking the users' interest similarity as a basis to judge the user similarity.(2)Considering the vulnerability of the old algorithm,the new one will take the trust and interest similarity among users into consideration to calculate the comprehensive similarity,so the inaccurate recommendation results caused by fake rating will be eased.(3)Propose a collaborative filtering recommendation algorithm which bases on the best trust path.The method to getting the average by the multipath of the old algorithm will be replaced by the best trust path.On the basis of the considering other users' prestige,the new algorithm will choose the best trust path from the multiple trust path to improve the problem that the old algorithm only cares prestige of ultimate users rather than the objectivity.The outcome of research indicates that comparing the collaborative filtering recommendation algorithm based on users and the collaborative filtering recommendation algorithm based on trust,the algorithm mentioned in this paper has the following advantages :(1)The accuracy of recommendation is higher.(2)The efficiency of operation is higher.The operation time of the algorithm mentioned in this paper is 1/4 of the fusion trust recommendation algorithm.When the trust path increases,The advantage of the algorithm mentioned in this paper is more obvious.