Blockchain-Based Credible and Privacy-Preserving QoS-Aware Web Service Recommendation

Abstract

With a growing number of alternative Web services that provide the same functionality, QoS-aware Web service recommendation is becoming increasingly important. However, collecting users’ observed QoS values is a challenging task for a recommender system. First, users don’t want to supply their observed QoS values due to privacy. Second, some user-contributed QoS values may be untrustworthy. There have been some centralized works on credible QoS prediction or privacy-preserving QoS Prediction. However, no research has been done to solve both the two problems simultaneously. Also, it’s difficult to guarantee the fairness and independence of the central server. In this paper, we propose a Blockchain-based Credible and Privacy-Preserving QoS-Aware Web service Recommendation framework. We first separate the traditional Matrix Factorization model into two disjoint parts: private factors and public factors, and train public factors collaboratively while keeping private factors secret. Then, we use blockchain, which based on the peer-to-peer network, to implement our proposed model. Through blockchain, users who don’t trust each other can reach a consensus without a central server. We conduct a series of experiments on a realworld dataset and analyze the proposed scheme in terms of accuracy, privacy, security, and complexity.