Hybrid Multi-Agent System for Metalearning in Data Mining

Abstract

In this paper, a multi-agent system for metalearning in the data mining domain is presented. The system provides a user with intelligent features, such as recommendation of suitable data mining techniques for a new dataset, parameter tuning of such techniques, and building up a metaknowledge base.

The architecture of the system, together with different user scenarios, and the way they are handled by the system, are described.

Keywords

• Artificial intelligence
• Data mining
• Multi agent systems
• Meta-knowledge
• Metalearning
• Parameter-tuning