Personalization has been named as one of the key challenges in information retrieval. In a world where search results have been optimized for several years already to serve the mass of all users, the last steps to go to yield considerable improvements in search quality are widely believed to need as one building block personalization. Personalization has already been tackled in various retrieval settings: in Web search, XML search, and in databases. The scenario we consider is a mixture of these settings. We study the problem of personalizing searching a knowledge-base which bears commonalities with databases as the information searched is rather structured without free text. A concrete methodology will be presented for the creation of a user profile this setting.