K-means Clustering Algorithm For Differential Privacy Preserving And Its Application In Recommendation System

In the era of big data,the first important thing to consider is how to use big data correctly and reasonably to bring convenience to daily life,and at the same time,the problem of information leakage needs to be considered.The recommendation system has solved the problem of information overload caused by big data to a certain extent,and differential privacy technology can achieve the purpose of information protection.This dissertation designs differential privacy preserving K-means clustering algorithm,which use differential privacy technology to solve the problem of user privacy leakage in K-means clustering.In addition,this paper combines differential privacy preserving K-means clustering algorithm with RBM algorithm to design a recommendation system to solve the problem of information overload in the era of big data.This dissertation investigates the existing k-means clustering algorithm for differential privacy and the corresponding recommendation algorithm,and focusing on two aspects.On the one hand,how to apply differential privacy technology to K-means clustering to protect user privacy attributes while also ensuring the availability of clustering results;on the other hand,how to apply differential privacy preserving K-means clustering algorithm and RBM Combined to design a new recommendation algorithm.the main research work includes:(1)Aiming at the leakage of user privacy information in the K-means clustering algorithm,this paper proposes an efficient differential privacy K-means clustering algorithm by clustering merge and adaptively adding noise.The design idea is: firstly,select more data points than the specified number of clusters in the dataset as the initial cluster centroids;then,add adaptive noise in the process of optimizing the centroids each iteration;Finally,after the cluster is stabilized,multiple clusters are merged into a specified number of clusters.(2)Aiming at the problem of information overload in the context of big data,this paper designs a recommendation algorithm combining differential privacy protection K-means clustering and RBM.The specific idea is as follows: Firstly,the data points in the data set are divided into different sub-classes through the K-means clustering algorithm for differential privacy protection;then,each cluster is individually generated a recommendation model according to the RBM algorithm.When a user has a recommendation requirement,first determine the cluster to which the user belongs,and then predict the user's behavior based on the recommendation model corresponding to the cluster to obtain some items that the user may be interested in.