Model Checking Commitment Protocols Mohamed El-Menshawy, Jamal Bentahar, and Rachida Dssouli Concordia University, Faculty of Engineering and Computer Science, Canada m_elme@encs.concordia.ca, bentahar@ciise.concordia.ca, dssouli@ece.concordia.ca Abstract. We investigate the problem of verifying commitment proto- cols that are widely used to regulate interactions among cognitive agents by means of model checking. We present a new logic-based language to specify commitment protocols, which is derived from extending CTL * with modalities for social commitments and associated actions. We re- port on the implementation of the NetBill protocol—a motivated and specified example in the proposed language—using three model check- ers (MCMAS, NuSMV, and CWB-NC) and compare the experimental results obtained. 1 Introduction A society mainly evolves through communication among cognitive participat- ing entities. In such a society, people interact and exchange information with each other to satisfy their goals. Such communication requires languages and mechanisms to structure interactions among participants within dialogues. Correspondingly, a cognitive agent-based model for an artificial society should provide adequate support for communication languages by providing means to specify interaction protocols that regulate intelligent interactions among agents. It is also beneficial to equip this model with a formal verification method to eliminate errors in the sense that protocols comply with the given specifications and to increase confidence on the model in terms of safety and efficiency. Motivation. This paper introduces a new logic-based language to specify a spe- cial kind of interaction protocols, called commitment protocols [5,6,13,12]. This logic extends CTL ∗ [4] with modalities for social commitments and associated actions. Social commitments usually represent the contractual obligations that associate the cognitive agents with one another in order to capture multi-agent intelligent interactions in this artificial society. In this context, social commit- ments model cognitive aspects of agent interactions by enforcing them to keep the fulfillment of their contracts and reducing cognitive dissonance in agent be- haviors. Moreover, we adopt these commitment protocols as they are increasingly used in different applications such as modeling business processes [5,6,13] and de- veloping web-based applications [11,2,7]. Conventionally, commitment protocols are defined in terms of creation and manipulation of social commitments that capture a rich variety of interactions among agents. Several approaches have K.G. Mehrotra et al. (Eds.): IEA/AIE 2011, Part II, LNAI 6704, pp. 37–47, 2011. c  Springer-Verlag Berlin Heidelberg 2011