The establishment of partnerships to create virtual organizations: A multiagent approach

Virtual Organizations are aggregations of independent organizations or individuals aiming to contribute to a common goal. Two basic steps are needed to assemble a virtual organization: (i) a business process partitioning, and (ii) a partners' selection process. Broadly speaking, in this paper we present a computational approach to model organizations based on Distributed Artificial Intelligence - Multiagent Systems (DAI-MAS) as well as Symbolic Learning (SL) paradigms. Each organization, which is seen as an agent, is provided with the needed observation, planning, coordination, execution, communication and learning capabilities to perform its social roles. In particular, we present a specific inter-organization relationship: the selection process that leads to the automatic establishment of contracts between organizations. This selection process is composed of a bid evaluation phase followed by a negotiation phase as a mean for agents conflicts resolution. Through negotiation interactions, a set of offer and counter-offer values which are seen as instances (positives and negatives) are supplied for further analysis in order to support the learning activities. The contribution of our work lies, not only, on the computational model proposed for the society of organizations, but also in some extent, on the learning methodologies applied to the established partnerships, in particular, and to the community, in general.