Domain Theory Verification Using Multi-agent Systems

We consider the use of multi-agent systems to verify a domain theory, which describes the object types and their properties, relationships and connections between them, as well as a set of true statements about the functioning laws of the domain objects. We study two approaches to the creation of multi-agent systems capable of verifying the compliance of types and object properties in the real domain, with types and properties defined in the theory. Requirements for the properties and relationship structure of intelligent agents are determined for each approach, as well as a mechanism for creating networks of multi-agent systems, which maximize efficiency and effectiveness in the verification of a given fragment of a domain theory. Simulation results show that the first of these approaches is considerably more efficient in terms of total verification time of a domain theory.