Towards to similarity identification to help in the agents' negotiation

Abstract

Enterprise delegates Agents' Negotiation is a simpler task if the enterprises involved in the transaction have homogeneous representation structures as well as the same domain of discourse, thus the use of a common ontology eases semantic problems. However, in real-life situations, real problems involve heterogeneity and different ontologies often developed by several persons and tools. Moreover, domain evolution, or changes in the conceptualisation might cause modifications on the previous ontologies once there is no formal mapping between high-level ontologies. We are proposing a method to be used by an Ontology-Services Agent to make Agents to understand each other despite their different ontologies. The method uses the natural language description of each involved item/product/service and combining statistical, clustering and suffix stripping algorithms finds out similarities between different concepts represented in different ontologies. © Springer-Verlag 2004.