Multimedia contents are accessible on a wide variety of devices, such as laptops, tablets and smartphones. The heterogeneity of such devices requires adaptation of these contents. Since the last decade, a fair amount of research has been conducted in order to develop adaptation frameworks. Most of them are exploiting profiles in order to guide adaptation processes. In some previous publications, we have proposed a model for specifying rich profiles containing complex constraints expressions. However, profiling information is currently mainly based on quantitative values. For example, if my battery level is less than 15%, do not play videos. In this paper, we improve our previous proposal by adding qualitative information to such profiles. For that purpose, we have defined a multi-level approach for specifying qualitative aspects and associating them to quantitative values. These terms will be used in our profiles for specifying rich semantic constraints, such as if my battery level is "low", do not play "high-quality" videos. We have proposed a meta-model that allows specifying these kind of constraints and have used Semantic Web technologies in order to serialize and query our profiles.