An efficient privacy-preserving friendship-based recommendation system

With the development of the internet, recommendation systems play a significant role for providing personalised services in our life. However, this raises serious concerns about privacy since the system collects a lot of personal information. Thus, plenty of schemes have been proposed to address the privacy issues by using cryptographic techniques. However, with the rapidly increasing numbers of users and items, most of existing cryptography-based schemes become inefficient because of the huge computation cost. In this paper, we propose an efficient privacy-preserving scheme for recommendation systems. Compared with existing schemes, our scheme does not require that friends of user are online during computing predicted rating. Finally, we evaluate the performance of our scheme with the MovieLens 20 m dataset and it shows that our scheme can reduce the overhead of computation and communication.