Modeling and Multi-Agent Specification of IF-based Distributed Goal Ontologies Nacima Mellal, Richard Dapoigny, Patrick Barlatier, Laurent Foulloy LISTIC-ESIA, BP 806 University of Savoie 74016 Annecy, France {nacima.mellal, richard.dapoigny, patrick.barlatier, Laurent.foulloy}@univ-savoie.fr Abstract. The concept of service is central in the design of distributed systems. In this approach for example, the web is developing web services and grid services. Nowadays, it is essential to take into account the crucial aspects of the dynamic services, that is to say their ability to adapt and to be composed in order to complete their task. To this end, the first part of the present paper aims to describe the implementation of a methodology which deals the automatic composition of services in distributed systems. Each service is related to a goal and is represented by a functional model called an Ontology. The model relies on a core reasoning process between interacting functional components of the complex system following the Information Flow (IF) approach. Afterwards, in the second part, we propose an algorithm describing the mechanism of the dynamic composition, basing on the first part and using Multi Agent System (MAS), where the agents support the functional components of the complex systems. 1 Introduction Embedded computing systems must offer autonomous capabilities in the often hostile environment in which they operate. Moreover, these systems are becoming more and more complex such as distributed intelligent components which are able to communicate and to reason about actions. Such systems include command and control systems, industrial process-control plants, automotive systems, data- acquisition and measurement systems, just to name a few. Although these systems are dedicated to design-defined tasks, they must offer some level of intelligence to cope with the dynamic and unpredictable environment with which they interact. A computational model powerful enough to capture the intensional knowledge within distributed systems and expressive enough to serve as a basis for reasoning and planning is required. Upon the teleological assumption a formal model based on goal structures has been proposed [4]. These structures are elaborated with the mechanisms of Formal Concept Analysis (FCA) during the design phase and results in domain ontologies able to be coordinated at run-time within an Information Flow (IF)-based framework. Web Services foster an environment where complex services that are