The use of machine-generated ontologies in dynamic information seeking

Abstract

Information seeking is the process in which human beings recourse to information resources in order to increase their level of knowledge with respect to their goals. In this paper we offer a methodology for automating the evolution of ontologies and share the results of our experiments in supporting a user in seeking information using interactive systems. The main conclusion of our experiments is that if one narrows down the scope of the domain, ontologies can be extracted with a very high level of precision (more than 90% in some cases). The paper is a step in providing theoretical, as well as practical, foundation for automatic ontology generation. It is our belief that such a process would allow the creation of flexible tools to manage metadata, either as an aid to a designer or as an independent system ("smart agent") for time critical missions. © Springer-Verlag Berlin Heidelberg 2001.