Stojanovic, NenadDadam, PeterReichert, Manfred2019-10-112019-10-1120043-88579-380-6https://dl.gi.de/handle/20.500.12116/28752In this paper we present a novel approach for discovering preferences of a user who searches for information in an ontology-based portal. The approach is based on analysing the user's searching behaviour regarding the given ontology and the content of the underlying information repository. In that way we develop so called short-term user's profile, that can be used for recommending relevant information to the user. The approach is realized as a method for ranking query refinements in the so-called step-by-step query refinement process. It is very suitable for modelling interactive searching in digital libraries or product catalogues.enText/Conference Paper1617-5468