Towards Con-Resistant Trust Models for Distributed Agent Systems

Amirali Salehi-Abari, Tony White
2009 International Joint Conference on Artificial Intelligence  
Artificial societies -distributed systems of autonomous agents -are becoming increasingly important in e-commerce. Agents base their decisions on trust and reputation in ways analogous to human societies. Many different definitions for trust and reputation have been proposed that incorporate many sources of information; however, system designs have tended to focus much of their attention on direct interactions. Furthermore, trust updating schemes for direct interactions have tended to uncouple
more » ... pdates for positive and negative feedback. Consequently, behaviour in which cycles of positive feedback followed by a single negative feedback results in untrustworthy agents remaining undetected. This con-man style of behaviour is formally described and desirable characteristics of con-resistant trust schemes proposed. A conresistant scheme is proposed and compared with FIRE, Regret and Yu and Singh's model [Yu and Singh, 2000]. Simulation experiments demonstrate the utility of the con-resistant scheme.
dblp:conf/ijcai/Salehi-AbariW09 fatcat:6wjwcxosrjb2zj55cqbkp6hfeq