An endorsement-based trust bootstrapping approach for newcomer cloud services

This paper addresses the challenge of providing trustworthy recommendations on newly deployed cloud services/resources for which little or no evidence about their trustworthiness is available. We also provide a two-level dishonesty discouragement mechanism to fight against unfair recommendations at both the collection and aggregation levels. Our solution consists of a (1) mechanism to allow users to self-assess the accuracy of their recommendations and autonomously decide on whether to participate in the recommendation process or not, (2) machine learning technique that generates reliable endorsements on newcomer items through extracting hidden similarities among the specifications of new and existing ones, (3) dishonesty-aware aggregation technique for endorsements coming from multiple advisors, (4) credibility update mechanism that captures the dynamism in the endorsers’ credibility, and (5) incentive mechanism to motivate advisors to participate in the endorsement process. Experiments conducted on the CloudHarmony and Epinions datasets show that our solution improves the accuracy of classifying newly deployed cloud services and yields better performance in protecting the recommendation process against Sybil attacks, in comparison with four existing recommendation approaches.