In companies a large amount of information is maintained that is accessible via network communication tools. This makes searching for a particular piece of information a di cult task. We identi ed two major di culties that are experienced using existing search engines. These di culties are that many of the existing search engines only take syntactic informationinto account and therefore do not nd allrelevant documents, and that many of the retrieved documents do not contain relevant information. In this paper we present a search engine that tackles both of these problems. The system, Dwebic, is based on a description logic knowledge base that deals with strict taxonomic information as well as with default information. Dwebic has a world-wide web interface, but our approach does not depend on the world-wide web.