Risk Assessment and Trust in Services Computing: Applications and Experience

Service-orientation is effective at managing complexity and dynamicity at a programmatic level, but there is still much work to be done in understanding and improving the trust that users place in a system's outputs, and the extent to which they understand the associated risks of decisions recommended by a system. This is crucial if we are to improve the uptake and real-world effectiveness of service-based decision-support systems whilst also reducing the risks (both perceived and actual) of using such systems. This paper presents the current progress of the STRAPP project, which is designing and engineering novel trust and risk assessment mechanisms for services computing and applying these to a number of real-world service-based decision-support systems. A new layered architecture model for trust and risk is introduced and described in detail, and we present our state-of-the-art work in risk-assessment, demonstrating the relationship between provenance data and risk via a mathematical model. We then give a detailed description of our latest software demonstrator, integrated into the Rolls-Royce Equipment Health Management system, and comprehensively discuss the lessons we have learnt from developing such a complex, holistic, and applied real-world system. Finally, we describe the future work we plan to complete.