Research On The Fuzzy C-means Clustering Recommendation Technology With Differential Privacy Protection

With the rapid development of Internet and the expansion of network information,users are helpless to face a large number of commodity information,and businesses can not get to know the needs of users in time and accurately and recommend goods for them.In this context,collaborative filtering recommends timely personalized recommendation according to user preferences,and is not dependent on user information or commodity attributes.It has become a widely used recommendation technology in e-commerce websites.However,with the explosive growth of data volume,the amount of computation of traditional collaborative filtering recommendation algorithm has also greatly increased,which has affected the efficiency of computation.The traditional method to solve the problem of oversize calculation is to introduce clustering algorithm,but one data in traditional clustering algorithm can only belong to one cluster,while one data in reality may have similarities in different aspects of data in multiple clusters,so the hard clustering of traditional clustering algorithm may affect the accuracy of recommendation.Moreover,due to the risk of privacy information leakage in the recommendation process,some users are reluctant to disclose privacy information,which greatly reduces the quality of recommendation.In view of the above situation,the main work of this paper is as follows:1.Aiming at the hard clustering problem of traditional clustering algorithm,the improved fuzzy C-means clustering algorithm is used to cluster users.The IKA algorithm is introduced to initialize the clustering center to make the cluster center evenly distributed.The membership matrix obtained by fuzzy C-means clustering algorithm finds nearest neighbors in the category of the target users,and reduces the computation range of nearest neighbors.It solves the hard clustering problem and the local optimum problem of traditional clustering algorithm.2.In order to protect the user's privacy information and guarantee the quality of the recommendation,after in-depth analysis of the privacy disclosure of the recommendation process,the Laplace noise is introduced into the center of the fuzzy C-means clustering process and the near neighbor of the target user's K.The security problem of the recommendation process is solved.3.In order to improve the accuracy of recommendation,the recommendationalgorithm is improved.Based on project category,a filling algorithm of scoring matrix and a similarity improvement algorithm based on user interest are introduced.Finally,in order to verify the effectiveness of the new algorithm proposed in this paper,the new algorithm is compared with other typical algorithms,and the effectiveness of the new algorithm is proved by the analysis and comparison of the experimental results.4.Design and implement a movie recommendation system with fuzzy C-means clustering algorithm with differential privacy protection,and analyze in detail from requirements analysis,architecture design,database design and so on.Finally,the implementation of movie recommendation system is completed by using Myeclipse development platform and Mysql database.The experimental results show that the fuzzy C-means clustering algorithm with differential privacy protection can solve the hard clustering problem and local optimal problem in traditional clustering,improve the clustering effect,fill in the prediction matrix,improve the calculation of user similarity,solve the sparse problem of the scoring matrix and improve the push.Recommend the accuracy;introduce the noise that satisfies the differential privacy in the fuzzy clustering and recommendation process,while ensuring the recommendation quality while solving the security problem.A comprehensive analysis of the evaluation indexes shows that the new algorithm proposed in this paper has higher recommendation accuracy while ensuring the security of the recommendation system.