Taxonomy-Based Hybrid Recommendation System for Lifelong Learning to Improve Professional Skills

In light of the vast amount of information related to lifelong learning courses, this paper proposes a hybrid recommendation system to overcome information overload. This system is based on user taxonomy profiles and makes it possible to extract unstructured data from multiple sources. The use of taxonomy allows knowledge about both user profiles and course contents to be modeled, thus enhancing system performance. The recommender engine operates in four phases: the first step uses a collaborative filter to determine users' areas of work; another collaborative filter is then used on this result to determine the related skills of the profiles; a third content-based filter is applied to make a preliminary course shortlist; followed by a final recommendation that is refined by using heuristics. The proposed recommendation system was tested on 120 user profiles and an improvement in the quality of the recommendation was observed.