Abstract: Automatic discovery of information in educational data has been broadening its horizons, opening new opportunities to its application. An open wide area to explore is the recommendation of undergraduate programs to high school students. However, traditional recommendation systems, based on collaborative filtering, require the existence of both a large number of items and users, which in this context are too small to guarantee reasonable levels of performance.In this paper, we propose an hybrid approach, combining collaborative filtering and a content-based architecture, while exploring the hierarchical information about programs organisation. This information is extracted from courses pro- grams, through natural language processing, and since pro- grams share some courses, we are able to present recommendations, not just based on the performance of students, but also on their interests and results in each of the courses that compose each program.