The aim of this paper is to show a context-driven use of automatic tools for medical ontology design. Following an analysis of the limitations of the content-driven and quantitative approaches currently used by knowledge organization specialists, we propose a contextualized and qualitative approach. This approach enables the definition of user-oriented resources and criteria for conducting semi-automatic terminology extraction and annotation tasks, ultimately supporting the development of a useful allergology ontology. At the same time, it reveals a paradox: although automatic tools are intended to reduce time-consuming processes, qualitative methods require considerable time and effort to collect and analyze contextual data.
