Norms in multiagent systems: From theory to practice

There is a wide agreement on the use of norms in order to specify the expected behaviour of agents in open MAS. However, current norm formalisms focus on the declarative (static) nature of norms. In order for norms to influence the actual behaviour of agents, these norms should be translated into operational representations, which can be used to constrain or direct actions of agents. In this paper we present our work on implementation of norm enforcement and issues on verifiability that highly affect this enforcement. We propose some mechanisms to be included in agent platforms in order to ease the implementation. Finally we discuss how different norms that apply for the same agent can be combined without resorting to default logic theorem provers. © 2005 CRL Publishing Ltd.