Role-Based Management and Matchmaking in Data-Mining Multi-Agent Systems

Abstrakt

We present an application of concepts of agent, role and group to the hybrid intelligence data-mining tasks. The computational MAS model is formalized in axioms of description logic.

Two key functionalities — matchmaking and correctness verification in the MAS — are provided by the role model together with reasoning techniques which are embodied in specific ontology agent. Apart from a simple computational MAS scenario, other configurations such as pre-processing, meta-learning, or ensemble methods are dealt with.

Klíčová slova

• MAS
• role-based models
• data-mining
• computational intelligence
• description logic
• matchmaking
• closed-world assumption