A Belief-Goal-Role Theory for Multiagent Systems

Agent-oriented techniques represent a promising approach for engineering complex systems where interaction is probably the most important single characteristic. Accordingly, the recent years have witnessed the emergence of different approaches for the study of intelligent agent-based systems. One such architecture views the system as a rational agent having certain mental attitudes of Belief, Desire and Intention (BDI). This paper explores a particular type of rational agent, a Belief-Goal-Role agent. Unlike most previous work, our approach has been to characterize the mental state of the agents that leads them to take part in cooperative action. Hence, beliefs, goals and roles are relevant to our study of cooperation which have lead to the identification of communication concepts (beliefs and goals) and organization concepts (roles). The model is formalized by expressing it as a theory in a first-order, multi-modal, and linear-time logic. We use labeled transition systems to deal with the truth conditions of the formulae of our theory as well as with the behavioral semantics of our agents. We illustrate our work with the well-known prey/predator game.

[1]  Yoav Shoham Agent-Oriented Programming , 1993, Artif. Intell..

[2]  Sergio Bandinelli,et al.  Supporting Cooperation in the SPADE-1 Environment , 1996, IEEE Trans. Software Eng..

[3]  Hector J. Levesque,et al.  Intention is Choice with Commitment , 1990, Artif. Intell..

[4]  Peter Wegner,et al.  Why interaction is more powerful than algorithms , 1997, CACM.

[5]  Jaakko Hintikka,et al.  Impossible possible worlds vindicated , 1975, J. Philos. Log..