A survey on blockchain-based Recommender Systems: Integration architecture and taxonomy

A Recommender System (RS) is an integral part of present-day leading web services, such as YouTube, Amazon, Netflix, and many others. Modern RSs are challenged to go beyond their traditional role of predicting user preferences to efficiently provide reliable, carefully personalized, and highly accurate recommendations. This paper thoroughly explores and analyzes state-of-the-art literature surveys on RS to extract important challenges and open issues. Our goal in this paper is to survey the literature to extract essential features of RSs and Blockchain (BC), focusing on their integration. Because of the lack of an existing foundation of BC-based RSs, the intrinsic BC aspects in RSs are identified and described. Integrating BC technology within RSs can achieve many benefits such as transparency, decentralization, and security. To that end, a thorough study of the papers on current BC-based RSs is presented along with a synthesized comprehensive taxonomy. Furthermore, a modular RS architecture, encompassing on-chain and off-chain storage and computation processes, is designed. This paper also includes a thorough discussion on the validity of the proposed architecture, BC limitations concerning RSs, and the derivation of a rich set of pointers to future research directions.