Lim Choi Keung, S.N. and Griffiths, N. (2008) Indirect Recommendations for Improved Trust Assessment. In: 11th International Workshop on Trust in Agent Societies (TRUST 2008), Estoril, Portugal.

Preview

PDF - Published Version Download (138Kb) | Preview

Abstract

Agents in open and dynamic environments face the challenge of uncertainty while interacting with others to achieve their goals. They face quick and unforseen changes to the behaviour of other agents and the population itself as agents join and leave at will. Since agents are assumed to be self-interested, it is essential for them to be able to choose the most reliable interaction partners to max- imise the success of their interactions. Efficient agent selection requires information about their behaviour in different situations. This information can be obtained from direct experience as well as from recommendations. This paper presents a trust and reputation model, which allows agents to select interaction partners efficiently by adapting quickly to a dynamic environment. Our approach is built upon a number of components from several existing models to assess trustworthiness from direct interactions and recommen- dations. We take a multi-dimensional approach to evaluate trust and reputation and include indirect recommendations as another source of trust. This reinforces our previous work on recommen- dation sharing, which includes information about the recency and relevance of interactions, allowing an evaluator to select recom- menders based on trust.

Actions (login required)