The semantic understanding of domain-specific terminology typical for communities, like a scientific conference, requires both, the analysis of daily spoken and written language in documents. A living taxonomy can help to cluster knowledge and classify questions against it, as well as to build up profiles of expertise. Text analysis modules support the parsing of training materials to extract a significant set of key phrases. However, the creation of a balanced categorisation tree requires the experience of community members. The Spree project introduces an OntoEditor$^{TM}$ to optimize automatic and manual processes to setup an expert finding \& communication application, following well-known Web 2.0 paradigms.