An Argument-Based Multi-agent System for Information Integration

In this paper we address the problem of obtaining a consolidated view of the knowledge that a community of information agents possesses in the form of private, possibly large, databases. Each agent in the community has independent sources of information and each database could contain information that is potentially inconsistent and incomplete, both by itself and/or in conjunction with some of the others. These characteristics make the consolidation difficult by traditional means. The idea of obtaining a single view is to provide a way of querying the resulting knowledge in a skeptical manner, i.e., receiving one answer that reflects the perception of the information community. Agents using the proposed system will be able to access multiple sources of knowledge represented in the form of deductive databases as if they were accessing a single one. One application of this schema is a novel architecture for decision-support systems (DSS) that will combine database technologies, specifically federated databases, which we will cast as information agents, with an argumentation-based framework.