An efficient privacy protection mechanism for recommendation using hybrid transformation technique

The rapid evolution of the internet led to a development of an effective tool for coping with information overload. Recommendation system is one of the most widely adopted and perceptible technologies for solving information overload issue. Collaborative filtering is the most popular technique used in recommender system. But it has several limitation such as sparsity, scalability and most vital of all privacy. As proposed by the researchers, the sparsity issue can be alleviated by incorporating the trust measure in the recommendation system. Customers mostly provide false information because of privacy breaches which in turn affect the accuracy of the recommendations. In this paper, a hybrid transformation technique is proposed which fuses Principal Component Analysis and Rotation Transformation (PCART) to protect users' privacy with accurate recommendations based on trust. The performance of our method is evaluated experimentally using MovieLens Dataset. Our experimental results shows, the hybrid transformation techniques provides better recommendations with ensuring privacy compared to existing approaches.