Social layers in agents' behavior evaluation system

Abstract

Behavior evaluation is an approach to a security problem in a multi-agent system that reflects security mechanisms in a human society. The main idea of this approach is behavior evaluation of all agents existing in society that is done autonomously by every agent belonging to that society. All autonomous behavior evaluations have to be collected and processed in order to create a collective decision of a society of agents. This approach reflects security mechanisms existing in a small society in which every human being has enough possibilities to observe and evaluate all other members of the society. This results in large computational complexity. In this paper a modification to behavior evaluation is presented which involves two simple social layers. Social layers are characteristic for more complex and larger societies and could be a means of lower computational complexity. © 2008 Springer-Verlag Berlin Heidelberg.