Trust as a form of defeasible reasoning

Abstract

Computational models of Trust have recently emerged as a way to exploit the human notion of trust in open and collaborative environments. This thesis provides a novel computational model of trust based on defeasible reasoning. Our starting point is an analysis of the nature of trust, that appears as a form of reasoning where arguments used are defeasible rather than deductive, subject to be defeated or strengthened in light of new evidence or conflicting arguments. The research issue investigated is the following; if trust is a form of defeasible reasoning, the application of defeasible reasoning computational techniques should improve the quality of trust computation.