Multiagent-based approach to distributed data mining

Data Mining and Knowledge Discovery in Databases (KDD) involves search for structure patterns or models in large and typically multidimensional data. The current generation of data mining tools has limited capacity and performance,,since these tools tend to be sequential. The aim of this paper is to integrate hardware with multiagent learning software approach to exploit parallel and distributed data mining. By combining techniques from distributed Al and machine learning, software agents equipped with learning algorithms mine local databases. These agents work in parallel and communicate with each other during the discovery process in order to integrate the knowledge obtained The paper explores the effect of distributing knowledge and learning capabilities by investigating different configurations of agents, several learning algorithms, knowledge representation etc.