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 trustworthiness 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.