Research On Task Collaboration Based On Multi-Agent In Data Mining System

This paper introduces data mining, multi-Agent and the background of the emergence of distributed data mining firstly, then introduces ontology, by which we discribes the concept model of multi-agent, task workflow, algorithms, operators, query and convertion method of ontologies and inference method of distributed data mining, on the basis of previous discussion, this paper describes the main algorithmic routine of the distribution of tasks, task optimization, task collaboration between agents and the combination of result produced by the work flow. At last it discusses concisely the system model and architecture of DDMS.On the basis of the theoretical exploration, this paper implements a demo DDMS system, which is built on the Agent developing architecture of Cougaar, a program imbursed by DARPA. In the demo system, the author experiment the critical algorithms discussed above.