Currently, we are living in the era of ubiquitous environments, that introduces the possibility to serve users in multiple context situations. Therefore, with this large variety of situations, diversity of preferences, and multiplicity of devices to respond to users’ needs, it becomes a necessity to have a solution that ensures the correct identification of specific situations. Otherwise, the identification process may be incomplete as it could be based beforehand on predefined situation rules, identified either by users or by the developers. Moreover, it is a cumbersome task for each user to identify his specific rules in any daily life cases. This emerges the need to recommend the right situation rule in the right context for the right person. Traditional recommender systems are mostly built to provide recommendations only based on user preferences, using a content or a collaborative approach and neglect situational context information, such as location, time and role. In this paper, we propose ontology-based dynamic context-aware recommendation system that enriches automatically profile’s situation rules in different domains (shopping, work, travel, etc.). It exploits users’ experiences by offering to the users a high level of comfort and a better-customized user experience. A new situation rules’ learning process is proposed to classify situation rules according to rule ontology before applying the recommendation process to achieve a high quality of recommendation considering three context categories (user preference, situation context, and device capability). To illustrate our approach, we have presented a case study that details how the whole process works with our recommendation system.