Rob Brennan, Karl Quinn, Declan O'Sullivan, Dave Lewis, Vincent P. WadeOn the Application of Paired Comparison to TrustEricsson R&D Ireland, Trinity College Dublin (Ireland)This paper describes a flexible approach to the internal representation, tracking and inference of trust in ubiquitous computing environments in the presence of heterogeneous trust models. Our method is based on knowledge engineering techniques and the application of modern paired comparison statistical methods from operations research.
Mechanisms for the estimation and propagation of trust through the Semantic Web, and previously electronic commerce or other secure relationships, have been for many years a subject of intense debate. Despite this dialog there is still little agreement on fundamental issues such as the properties of trust as a concept, the types of operations that may be usefully applied to a trust value, or even what kind of domain trust values could or should be drawn from. Nonetheless trust has been adopted by many as a useful concept for a wide variety of communications network applications. For example, the Semantic Web research community has embraced it enthusiastically.
In this paper we present some initial ideas on how current developments in operations research and knowledge engineering may be applied to the problem of representing and propagating trust in ubiquitous computing environments where multiple, inconsistent trust models are likely to hold sway for the foreseeable future.
Our role here is not to define a full formal model for trust; it is to define a tractable solution for those parts of the concept of trust between software entities that may find immediate applicability in our problem domain; the dynamic selection and composition of (web) services from atomic service components in ubiquitous computing environments.
|
