A new knowledge engineering research paradigm is emerging based on the shift from the construction of stand-alone knowledge/data based systems (possibly integrating different problem solvers) to the integration of multiple, at least semi-autonomous and in general heterogeneous software agents into the MAS architectural framework, allowing them to interoperate and thus cooperate within common application areas. Within these multi-agent systems even the users can be seen as fully integrated (human) agents, to which the software ones offer services tailored to their specific professional requirements. Such a vision entails bridging the different "views of the world" of knowledgeable agents through the commitment to common definitions of the conceptual entities and of the technical terms employed in knowledge/data bases.
We claim that within a multi-agent framework, highly empowered with respect to the single-system and the distributed knowledge/data approaches, ontological libraries and also standardized terminological repositories should be agentified into ontology and terminology servers providing other agents with the common semantic foundation required for effective interoperation. Application ontologies configured from library theories can thus be used to model application agencies shaping a particular view on the services provided by the MAS, such as those of multiple DBMSs and KBSs. Depending on the extent to which application ontologies overlap, or on the relationships that can be established among their definitions starting e.g. from information in a common originating ontological (sub)library, they can also be used by the ontology server to mediate (in a partial, semantically founded fashion) transactions among different application agencies.
This research effort is just at its earliest development stages, as there are many issues still left open. Actually one limitation shown by the application agencies implemented [Lanzola et al., 1995; Lanzola et al., 1996] is represented by the adoption of a fixed schema for connecting their component agents. So far, each agent knows in advance which are the other ones cooperating with itself and their exact location on the network. So the whole agency fails if any component for some reason becomes unavailable at a given time. The institution, within the ontology server, of a database storing information on ontology-agent pairs will ease the search of alternative candidates for the delivery of the requested service.