COMMUNICATION IN A MULTI-AGENT SYSTEM BASED ON TRANSPARENT INTENSIONAL LOGIC

The goal of this paper is to introduce Transparent Intensional Logic (TIL) as a framework apt for the specification of ontologies, knowledge representation and communication by messaging in a multi-agent system. We aim in particular at the specification of ontologies and their application in building and maintaining knowledge bases of intelligent agents. Since the intelligent agents should be able to learn by experience and deduce derived knowledge from their explicit knowledge in order to react adequately even in unexpected situations, we need a highly expressive logical framework to meet these goals. We show that TIL is such an expressive logic that makes it possible to make all the semantically salient features of a language explicit and logically tractable. Due to the ramified hierarchy of types we distinguish in TIL between a mode of the presentation of a function and the function itself and between a function and its value. Thus the agents can reason about concepts themselves, learn new compound concepts via refinement of less complex concepts and exhibit an adequate dynamic behaviour.