Business Intelligence Fusion Based on Multi-agent and Complex Network

Abstract

A fusion method of heterogeneous business intelligence (BI) technologies is put forward, named agent-network. This method treats BI system as a complex network composed of agents as its nodes. One agent is a unit of intelligence resource (IR) representing a computing model or an algorithm. A BI technology is a group of agents. Three basic mechanisms are discussed in detail. The IR aggregating and optimizing mechanisms can improve BI software to be a dynamical and flexible system. New technologies can be added into BI software continuously and less value existing technologies can be deleted from it. The IR using mechanism can always let the BI system to select an or a group of optimal technology (technologies) to respond to every specific user request by using some marketing mechanism such as negotiation, bidding and auction. The IR optimizing mechanism can keep the best agent and delete inferior agent by the performance of agents. The BI architecture is proposed based on our method. Different enterprises can customize their own BI service at a lower cost by using our method. In the future, we will develop a prototype software system based on our agent-network method to improve the decision level of enterprise.