Towards ontological reconciliation for agents

This paper addresses issues faced by agents operating in large-scale multi-cultural environments. We argue for systems that are tolerant of heterogeneity. The discussion is illustrated with a running example of researching and comparing university web sites, which is a realistic scenario representative of many current knowledge management tasks that would benefit from agent assistance. We discuss efforts of the Intelligent Agent Laboratory toward designing such tolerant systems, giving a detailed presentation of the results of several implementations.