This study is part of our ongoing research on the field of conceptual query expansion, currently focusing on a user-centred approach relating to whether non-expert users could directly interact with various types of KOSs 1 . In previous phases, we studied the initial query formulation for given information seeking tasks. We mapped, both lexically and semantically, the terms used to formulate a query to terms included in certain KOSs, namely the EUROVOC 2 and GEMET 3 thesauri and WIKIPEDIA 4 Building on the initial query formulation results, we explored the reformulations of the initial query, too. The hypothesis was that the outcome of this study would elaborate on the question whether non-expert users would eventually use terms included in the KOSs and are, thus, familiar with them constituting the KOSs candidate tools for conceptual query expansion in direct interaction with the users. . Lexical mapping of terms, which is under constant research [1], gave poor results, meaning that, in average, the terms were matched only for 49.7%, 41.4% and 27.4% of the cases respectively. We concluded that, concerning the initial query formulation, it is highly unlikely for non-expert users to use terms mapped to a domain-specific thesaurus such as GEMET. Considering, however, the high performance of its semantic relatedness (57.7%, 62.2% and 14.6% respectively for the three KOSs) we were led to believe that a domain-specific thesaurus is a good choice for bridging the semantic gap between the users' input and the terms used to represent certain content.