Feature selection by using privacy-preserving of recommendation systems based on collaborative filtering and mutual trust in social networks

Given the increasing growth of the Web and consequently the growth of e-commerce, the amount of data which users face are increasing day by day. Therefore, one of the key issues in today’s world is the extraction of knowledge from a large database. The recommendation systems are able to extract useful information from large databases. The information extracted by the recommendation systems may breach the privacy-preserving of individuals and increase the error rate. Concerns will grow along with the increasing privacy breaches, which are done by recommendation systems. In recent years, researchers have provided a variety of techniques for privacy-preserving and reduced error rates in recommendation systems. But most of these methods have not offered good solutions for privacy-preserving issues and reducing error rates. The aim of the proposed method is to provide a solution for users’ security concerns in common filtering systems with reduced error rates and more privacy preservation. In this article, we propose a privacy-preserving method for recommendation systems called PRS, which first uses an anonymous method to convert secondary data without user identification information. The existing trust data are measured in terms of resemblance and trust-weighted criterion and then converted from perturbation-based chaos to confidential data. Finally, these two algorithms have been used for clustering the data: fuzzy c-ordered means and particle swarm optimization. The results of experiments have been compared with state-of-the-art methods, which show the superiority of the proposed method in terms of classification error rates and privacy-preserving.