Interaction Modal Logic for multiagent systems based on BDI architecture Matías Alvarado, Leonid Sheremetov Center for Computing Research of the National Technical University, (CIC-IPN), Mexico Av. Juan de Dios Batiz esq. Othon de Mendizabal s/n. Col. Nueva Industrial Vallejo, México, D.F., C.P. 07738 matias@cic.ipn.mx, sher@cic.ipn.mx Abstract. The purpose of this paper is to develop a conceptual model of multiagent system composed of a set of BDI agents based on role modeling and modal logic of interaction. Agents perform actions taking roles within a scenario that obeys a plan conformed for a set of goals and establishes relations between the participants. These plans and relations define interactions between agents. All interactions are divided in communicative and physical ones; a common model of interactions of the both types is proposed for its further formalization as a modal logic of interaction. The key characteristics of this logic are dynamic reasoning and dynamic plan executions modeled through modal categories. 1. Introduction One area of much interest in DAI is the use of mathematical logic for specifying properties of agents and multiagent systems (MAS). One of the first works in this field was the Cohen-Levesque theory of intention [5]. To express this theory, the authors developed a quantified multi-modal logic, with modalities for representing beliefs and goals, and an apparatus for representing actions that was loosely based on dynamic logic [10]. Beliefs and goals were characterized using possible worlds semantics. Building largely on this work, attempts have been made to use similar logics to capture various other properties of agents [19] and multi-agent systems [17]. Rao and Georgeff's BDI theory [19], probably is most popular among the formalisms capturing exactly this balance. In this framework, based on Bratman's model of practical reasoning and CTL temporal logic proposed by Clarke and Emerson [3, 4], beliefs, desires, and intentions are introduced as the independent modalities. Nevertheless, these models were mainly models of BDI agents and not models of systems of rational agents. These systems usually consist of subsystems, subsystem components, interactions and organizational relationships. The interplay between subsystems and between their constituent components (agents) is most naturally viewed in terms of high level social interactions. This view and level of abstraction accords precisely with the treatment of interaction afforded by the MAS, described in terms of "cooperating to achieve common objectives'', "coordinating their actions'' or "negotiating to resolve conflicts'' [11]. Since agents make decisions about the nature and scope of interactions at run time, it is imperative that this key shaping factor is