Accountability in Service-Oriented Architecture: Computing with Reasoning and Reputation

Abstract

Accountability benefits the service-oriented computing (SOC) with trust and error reasoning. In this research, we study an accountability model with 3-D approach: detect, diagnose, and defuse, to discover and eliminate the root cause of problems when violations of SLA (service level agreement) occur in business processes. Our basic approaches are: (1) leveraging Bayesian network to diagnose the root cause of problems when uncertainty exists in business processes; (2) a continuous knowledge learning process to deal with the dynamic nature of SOC: the feedbacks about problematic web services are counted into services' reputation, and will impact future service selections and Bayesian network parameters learning. The performance study shows that our accountability model can rule out the root cause of problems in the whole business process effectively and efficiently with acceptable cost