The industrial districts are characterized by the agglomeration of medium and small-sized industries, localized within a certain geographic area with precise social and cultural connotations. A crucial element of industrial districts is the existence of a wide immaterial flow of knowledge and information. In this sense, the industrial districts seem to have a network shape, rather than a hierarchical one.

The paper aims to rationalize the transportation flow within industrial districts. In particular, it focuses on districts in the first stage of the evolution process; that is, when a lot of companies exist, with low specialized level and absent inter-firm cooperation. Therefore, the considered industrial districts are characterized by non-linear mechanisms of interaction among a variety of entrepreneurs, each of them trying to get his target by conflicting with others, increasing the complexity of the system under examination and making more difficult the strategies for material flows rationalization.

In this situation, a system able to facilitate both the contacts among the agents and the negotiation processes, that creates a community of agents in a district with rare or absent relationships among different companies represents the most adequate solution.

The paper deals with the definition of the community of agents in a real situation, using the communityware tools: web electronic mediums that facilitate the contact with collaborators who have similar interests and preferences, but do not know each other. To define the similarity concept, a fuzzy algorithm is proposed.