A market-oriented agent-based model for information retrieval

Abstract

Information Retrieval in the World Wide Web (Web IR) is essential for a number of activities and it is an active domain of research and development. The main challenges concern the relevance of the results provided to users' queries and the performance regarding respond-time. On the other hand, agent-based market systems prove to be efficient for implementing e-commerce or B2B applications on the internet, thanks to inherent properties such as prominency of interactions, scalability, flexibility, interoperability, etc. Although the use of agents in other application domains is not yet widespread, the integration of mobile agents into market mechanisms bring clear and efficient solutions to Quality of Service issues encountered in most distributed applications and notably in Web IR systems. Mobility allows defining the seller – buyer model of interaction, where agents act on behalf of final users or devices providing re-sources, while the generic Market Place architecture provides an organizational setting for the matching of demands and offers. The paper shows how this framework applies to Web IR and provides experimental validation results from a Jade implementation.