A Framework for a Scalable Agent Architecture of Cooperating Heterogeneous Knowledge Sources

Interoperability amongst heterogeneous information sources continues to pose enormous challenges to the database, AI and other communities. While significant progress has been achieved in system, syntactic, and structural/schematic interoperability, comprehensive solutions to semantic interoperability remain elusive. We present and motivate the conceptual framework underlying SCOPES (Semiotic/Semantic Coordination of Parallel Exploration Spaces), a scalable agent architecture designed to support interoperable, autonomous and heterogeneous knowledge sources. It is posited that the traditional approach to semantics is insufficient to account for a variety of misinterpretations in a realistic social world. Agents are viewed as existing in a social world (of beliefs, expectations, commitments, etc) constrained by pragmatics (intentions, communication, etc) and with particular semantics (meanings, propositions, validity, etc) and syntactics (formal structures, language, data, deduction, etc). We then concentrate on semantic interoperability and how this is handled in SCOPES. Semantic reconciliation is viewed as a non-monotonic query-dependent that requires flexible interpretation of query context, and as a mechanism to coordinate knowledge elicitation while constructing the query context. We elaborate on the specific concepts needed to build this context, and briefly discuss the SCOPES’ algorithm for context construction.