An Investigation into Trust & Reputation for Agent-Based Virtual Organisations

Abstract

Trust is a prevalent concept in human society. In essence, it concerns our reliance on the actions of our peers, and the actions of other entities within our environment. For example, we may rely on our car starting in the morning to get to work on time, and on the actions of our fellow drivers, so that we may get there safely. For similar reasons, trust is becoming increasingly important in computing, as systems, such as the Grid, require computing resources to work together seamlessly, across organisational and geographical boundaries (Foster et al., 2001). In this context, the reliability of resources in one organisation cannot be assumed from the point of view of another. Moreover, certain resources may fail more often than others, and for this reason, we argue that software systems must be able to assess the reliability of different resources, so that they may choose which resources to rely upon. With this in mind, our goal here is to develop a mechanism by which software entities can automatically assess the trustworthiness of a given entity (the trustee). In achieving this goal, we have developed a probabilistic framework for assessing trust based on observations of a trustee's past behaviour. Such observations may be accounted for either when they are made directly by the assessing party (the truster), or by a third party (reputation source). In the latter case, our mechanism can cope with the possibility that third party information is unreliable, either because the sender is lying, or because it has a different world view. In this document, we present our framework, and show how it can be applied to cases in which a trustee's actions are represented as binary events; for example, a trustee may cooperate with the truster, or it may defect. We place our work in context, by showing how it constitutes part of a system for managing coalitions of agents, operating in a grid computing environment. We then give an empirical evaluation of our method, which shows that it outperforms the most similar system in the literature, in many important scenarios.

More information

Identifiers

Catalogue record

Export record