The work presented in this thesis is part of a broader issue of Adaptive Learning System (ALS). He is particularly interested in Dynamic Adaptive Hypermedia Systems (DAHS). The main purpose is to examine the advantage of the fine-grained approach to adapt the course in these systems. In fact, it deals with automatic generation of courses tailored to a particular student, based on a set of educational resources and according to their needs, preferences and prerequisites. Learning resources called learning objects are currently indexed using metadata standards and educational standards as SCORM and LOM. These learning objects are assembled and associated to form individual training with a hypermedia presentation. We propose a new approach for the learning object granularity based on size, content and the media used. This approach provides a breakdown of content in fragments, which may be in different forms such as Introduction, definition, example, comment, etc. Each of these fragments is represented by multimedia bricks like text, image, video, sound, etc. All these elements are used to improve adaptability, in content, presentation and navigation in a hypermedia learning environment. In summary, based on a set of basic and granular resources, their semantic descriptions and the learner profile, the system expected to be able to generate adequate courses (courses, exercises, etc.) with a Web browsing interface. In this sense we have established the first brick of DAHS, which we call Adaptive Learning System for the"C" Programming Language (ALS-CPL). This particular system aims first to implement our research in programming learning environments as part of a broader project entitled Adaptive Learning System for Programming Languages (ALS-PL). It also aims to apply particularly the granular content approach for adaptability.
