A Recommendation System Based on Knowledge Gap Identification in MOOCs Ecosystems
ResumoThe consolidation of recommendation systems in a big data era brings opportunities in different scenarios to customize methods that recommend data. In the scenarios of the Massive Open Online Courses (MOOCs) ecosystems, these recommenders mainly support students in choosing the best courses from the platforms. However, the expansion of course platforms and the scarcity of student data increases the difficulty in finding courses, or even part of courses, that fill a given knowledge gap. In this paper, we propose a recommendation system to support students in finding the best modules or courses in these ecosystems. First, topic modeling techniques were implemented with Non-negative Matrix Factorization (NMF) to find similarities between multiple MOOCs providers. Then, a content-based recommendation provides recommendations to a user interested in acquiring new knowledge, based on a history extraction on those platforms. We evaluate our approach through an experiment with real data collected in multiple MOOCs providers. In addition, by comparing the NMF approach with a baseline Latent Dirichlet Allocation (LDA) technique, we verify the model effectiveness and show that our system is useful to this context.
Palavras-chave: MOOCs ecosystems, recommendation system, topic modeling, nonnegative matrix factorization
CAMPOS, Rodrigo; DOS SANTOS, Rodrigo Pereira; OLIVEIRA, Jonice. A Recommendation System Based on Knowledge Gap Identification in MOOCs Ecosystems. In: SIMPÓSIO BRASILEIRO DE SISTEMAS DE INFORMAÇÃO (SBSI), 16. , 2020, Evento Online. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2020 . DOI: https://doi.org/10.5753/sbsi.2020.13746.