Confidence-Aware Reputation Bootstrapping in Composite Service Environments

We propose a novel reputation bootstrapping approach for both composite and atomic services in service-oriented environments. We consider multiple factors which may implicitly represent reputations of new services. Our approach does not rely on empirical assumptions. In contrast, we propose a data-driven method to determine how much a factor can represent service reputation. The reputation-related factors are modelled in a layer-based framework. This aims to quantitatively describe the importance of factors in reputation bootstrapping. Furthermore, we define confidence to represent how reliable the bootstrapped reputation of a new service is. We evaluate our approach based on a real-world dataset. The experimental results demonstrate the feasibility and outperformance of our approach.