Cooperative agents for information gathering

Abstract

The rapid growth of the network-centered (Internet and Intranet) computing environments requires a new design paradigm for information gathering systems. Typically, in these environments, information resources are dynamic, heterogeneous and distributed. Moreover, these computing environments are open, in the sense that information resources may join or disjoin at anytime.

Agent-based technology provides a promising design paradigm and has a growing appeal for designing cooperative distributed systems. This thesis develops a multi-agent architecture for cooperative information gathering systems (CIGS). These autonomous, goal-driven agents cooperatively assist different users to locate and retrieve information from distributed resources in an open environment. The system architecture is comprised of three tiers: at the front end, the User Agents interact with the users to fulfill their interests and preferences; at the back end, the Resource Agents access and capture the content and changes of the information resources; and at the middle tier, the Broker Agents facilitate cooperation among the agents.

Cooperation is an essential concept for information gathering in a multi-agent setting. In this thesis cooperation has been analyzed and instrumented to govern the agents' interaction in an information gathering domain. Various interaction strategies have been proposed at different levels of the agent's model. Through these strategies the agent might to choose to exhibit selfish, benevolent, or cooperative behavior based on the interaction style. The feasibility of the proposed system has been demonstrated by implementing a prototype. This implementation demonstrates how CIGS could be used to transparently locate and retrieve information from dynamic resources that might be distributed and heterogeneous.