Abstract
The explosion of data on the Internet leads to challenges in working with them. Semantic Web and ontology are required to address those problems. Nowadays, ontology plays more and more an important role as a means in domain knowledge representation.In recent years, Non-Pharmacological Interventions (NPIs) have attracted a lot of attention in the health care community. NPIs can no longer stop at a professional discipline to describe them (psychotherapy, manual therapy, dietary supplement, adapted physical activity, e-health solution, etc.). It requires access to a more concrete level of description where each NPI can be evaluated by science, monitored by professionals and explained to the patient. To do this, an international and evolutionary classification based on the results of science is necessary. Thus, developing an ontology for NPIs is crucial. This ontology will facilitate bibliographic research, usage statistics and the identification of good practices.Constructing this ontology manually is time consuming and thus an expensive process. Particularly, the step of collecting the NPI terminology requires much more time than the rest, because of heterogeneous and big resources in the context of NPIs. An automatic or semi-automatic method is thus essential to support NPI experts in this task.Besides, ontologies are often complex with lots of classes, properties and relationships. They are not easy to understand by domain experts. Therefore, a simple and friendly visualization of the ontology for NPI experts needs to be considered. The visualization does not only help NPI experts to easily understand the ontology but also provides support for the NPI ontology development process.In this thesis, we propose methodologies to address the aforementioned challenges. The first contribution concerns the semi-automatic process for collecting NPI terms. Two approaches, knowledge-based and corpus-based, are presented to retrieve candidate NPI terms. A new similarity measure for NPI is proposed and evaluated. The second contribution is a new method for ontology visualization based on MindMap. This work aims at providing a simple and friendly tool to visualize an ontology which is used by domain experts. We propose a MindMap-based notation for ontology visualization by transforming ontology components to MindMap elements. A web-based tool is then implemented to convert OWL ontologies to FreeMind documents which can be imported by existing Mind-Mapping applications to make visualizations.