The discovery mechanisms of Web Services in the UDDI registries are currently quite weak. UDDI provides an interface for keyword and taxonomy based search. The problem is that it has to exist an exact keyword matching. If the consumer searches for the keyword “apartment”, but the provider registered her/his service under “flat”, the result will not be got. Additionally the existing taxonomies might not be appropriate enough, converging to very long result lists that do not really help. The main inhibitor is the lack of semantics. There are some attempts in improving the discovery situation and XML Topic Maps, as a representative of the Semantic Web, could help solving this problem. The present work addresses the stated issue and is, to my knowledge, a new solution. It aims to show how XML Topic Maps can be used as a tool for organizing and representing knowledge out of a specific domain, in our case bioinformatics. This additional semantic layer, expressed in XML Topic Map notation, will bring some benefits to a potential service requestor, who needs to discover the location of an unknown Web Service or resource in general. A prototypical topic map, representing a subdomain of the Gene Ontology, will be developed and with the help of an implementation it will be shown how the knowledge representation can be linked to UDDI. Finally it will be shown how the XTM can be navigated in different ways, in order to point the aim and find a previously unknown resource. XTM will actually be used for building a complete knowledge base; to each term which a potential user may want to navigate through, it will be assigned a number of external information resources. Some aspects have to be considered; the most important – due to the continuous evolution of the Gene Ontology, the XTM document has to be permanently maintained. At the end it was inferred as conceivable referring directly to the location of the WSDL file and doing the job totally without UDDI.