As Web services increasingly become important in distributed computing, some of the flaws and limitations of this technology become more and more obvious. One of this flaws is the discovery of Web services through common methods. Research has been pursued in the field of "Semantic Web services". This research is driven by the idea, to describe the functionality of Web services as accurately as possible and to create programs automatically out of already existing Web services. In this paper we discuss a new method for discovery and analysis of Web services. Our approach uses a Vector Space Search Engine to index descriptions of already composed services. Rather than generating or automatically composing applications, this approach provides developers with a valuable utility to browse repositories based on already existing information. Furthermore, we propose some additional modifications to extract the maximum amount of semantics from existing service definition repositories.