Scalable Privacy-Preserving Recommendation System

Abstract

Recommendation systems, which are used by e-business applications, play an essential role in our daily life. For example, companies like Amazon, Flipkart, YouTube use recommendation systems to generate customized recommendations from the user�s personal information. A personalized recommendation system�s primary purpose is to give users helpful suggestions on different things. The service provider has to access different kinds of user data in order to make suggestions, such as prior product buying history, demographic, and identifying details. Users, on the other hand, are wary about disclosing personal information since it may be readily abused by hostile third parties. To address this challenge, we propose a privacy-preserving recommendation system using homomorphic encryption, which is able to provide recommendations without revealing user rating information. Also, in this system, a service provider can use another service provider�s user rating database to improve the generated recommendation while protecting user�s privacy. The implementation of the proposed system on a publicly available database shows that the system is practical and achieves higher commendation accuracy.