Multi-agent knowledge acquisition and application in distributed enterprise database

In this paper the model of the distributed decision support system with task oriented learning function is proposed. This model is composed of local subsystems which are constructed based on the multi-agent technology to ensure its independence and cooperation ability. This structure can satisfy the local application and, by cooperating with other agents, it can also satisfy the global application. The task oriented learning function allows the system to obtain the needed knowledge to improve the decision-making ability. The knowledge discovery in database technique is a main tool for knowledge acquisition so that the system can explore useful rules from the enterprise database timely. Knowledge acquisition agents in different subsystems can exchange knowledge through the Internet and this improves the efficiency of knowledge acquisition. Knowledge valuation is considered in this model. Genetic algorithms, neural networks and graph theory based algorithms are also adopted to reinforce the knowledge base.