Finding a course or book on a specific subject in a directory can be tedious. The problem is even more accentuated by the multidisciplinary of some of these courses or books. Graduate students are responsible for choosing their study plan, the courses relevant to their field of research, but it is not obvious that they can make the right choice without needing to be guided or oriented. With a tool to establish the similarity between different documents, students could quickly find courses or books similar to those which, for one reason or another, are not available. To this end, several filtering systems have been proposed, but filtering based on content for the recommendation of courses or books, has so far not been addressed as done in this work, by resorting to the measure of similarity. based on Dice's coefficient, thus providing relatively accurate and comprehensive recommendations. The objective of this research is to propose a model allowing to establish the similarity between courses and books, while being based on their descriptions and on the calculation of their distance in a vector space . This reflection presents the content-based filtering system for recommending courses and books, providing suggestions based on their semantic similarity.