Building a fuzzy trust network in unsupervised multi-agent environments

Abstract

In automated and unsupervised multi-agent environments, where agents act on behalf of their stakeholders, the measurement and computation of trust is a key building block upon which all business interaction scenarios rely. In environments, where the individual and independent calculation of trustwor-thiness values for future negotiation partners is desired, flexible algorithms and models imitating human reasoning are crucial. This paper introduces a trust evaluation model that imitates human reasoning by using fuzzy logic concepts. Furthermore, post-interaction processes such as business interaction reviews and credibility adjustment are used to continuously build and refine an information repository for future trust evaluation processes. Fuzzy logic offers a mathematical approach encompassing uncertainty and tolerance of imprecise data, and combined with our highly customizable model, it allows to meet the security needs of different stakeholders. © Springer-Verlag Berlin Heidelberg 2005.