Providing recommendations for communities of learners in MOOCs ecosystems

Massive Open Online Courses (MOOCs) have been widely disseminated due to the arrival of Web 2.0. However, the growth of MOOCs brings some difficulties for students in choosing suitable courses in these ecosystems. In recent years, some recommendation systems emerged to solve this problem but remain limited since they do not identify the student’s prior knowledge broadly or the student’s goals. To overcome this limitation, this work proposes the Fragmented Recommendation for MOOCs Ecosystems (FReME), a recommendation system to suggest parts of courses from multiple providers (i.e., Khan Academy, Udemy, and edX). FReME is based on the student profile and on the MOOCs ecosystems perspective to balance the ecological environment and strengthen interactions. Moreover, we differ from the current recommendation systems since our method identifies and reduces the students’ knowledge gap optimizing the learning process. Experimental results conducted with a dataset integrating 3 MOOCs providers and 19 students demonstrated that the implemented techniques are more consistent than other approaches. Finally, it was verified through precision, utility, novelty, and confidence that our recommendations are 62,24% accurate, 68.89% useful, 72.81% reliable, and present new content in 99.12% of cases. These results validate that FReME supports students in reducing their knowledge gap. •A content-based recommendation using topic modeling and labeling techniques.•Recommendation of part of courses from multiple providers to support learners.•Implementation of a system with real data from providers, enabling the expansion.•The model’s number of topics and document labels are automatically defined.•Three quali-quantitative experiments confirm the system’s effectiveness.