A multiagent architecture for information retrieval on the World-Wide Web

The world-wide web or simply the web is a large distributed digital information space. The number of information resources available on the web is increasing rapidly. Although some useful information may exist somewhere on the web, locating such information by a web user often requires enormous search time and effort. To facilitate retrieving information from the web, many search tools and services have been recently developed. Though existing search tools are technically efficient, however, their retrieval effectiveness is very poor.

In this paper, we describe a multiagent architecture for effective information retrieval on the web. The modular nature of the architecture enables realizing incremental improvements in the performance of the retrieval system. Any of the several agents in the system can be replaced by a more effective agent, provided that the latter communicates using KQML (knowledge query and manipulation language) and is conversant with the domain ontologies. Retrieval effectiveness is achieved by employing user models, domain ontologies, and adaptive query reformulation via relevance feedback.