Research On Recommendation Algorithm Based On Trusted Neighbors

With the rapid development of social networks,the trust relationship between users plays an increasingly important role in the recommendation process.Studies have shown that,by considering trust information to calculate similarity,a better recommendation effect can be obtained.However,the existing trust-based recommendation algorithm ignores the following problems:(1)The algorithm does not take into account the impact of the "shilling attack" problem,resulting in poor system robustness and vulnerability to attacks.(2)The algorithm does not fully mine the score information of trusted neighbors in the trust network,so that there is still a serious problem of data sparseness in the direct trust calculation.(3)When there are multiple paths,the existing studies have ignored the trust weights of different paths,and have not fully tapped the trust value of longer paths,making the calculation of indirect trust not accurate enough.Aiming at the problem of shilling attack,by comparing shilling attack users and normal users,the behavior characteristics under the trusted network are different,multiple statistical characteristics are defined to detect shilling attack users.Simulation experiment results show that the algorithm can maintain a high detection rate and improve the robustness of the system;Considering the data sparsity problem of the trust network,the design idea of the collaborative filtering algorithm is adopted,and the score information of the trusted neighbor is added to the calculation of the user's direct trust degree.The algorithm makes users with higher similarity and higher similarity with trusted neighbors have higher similarity,and fully mine the information in the trusted network.Simulation experiment results show that the algorithm RMSE index has been significantly improved;In view of the shortcomings of the existing trust-based calculation model,the idea of collaborative filtering algorithm is adopted,and a path screening algorithm based on the difference of path items is proposed.The algorithm calculates the trust path between trusted users as a whole,so that the smaller the difference in the trust path score,the more the score items,the greater the trust value of the trust path with the shorter path length.At the same time,the mining of the trust information of the longer path.Simulation experiment results show that the algorithm has better recommendation accuracy and better real-time performance.