T7-ALT1-3 - LSBCTR: A Learning Style-Based Recommendation Algorithm

2. Research-to-Practice Full Paper
Thayron C. H. Moraes1, 2 , Itana Stiubiener1, Juliana C. Braga1, Edson P. Pimentel1
1 UFABC - Universidade Federal do ABC
2 UFMT - Universidade Federal do Mato Grosso

Abstract—This Research Full Paper presents a hybrid algorithm for the recommendation of Learning Objects (LO) aimed at students’ learning profiles. In this sense, the Learning Stylesbased Collaborative Topic Recommender (LSBCTR) algorithm was developed based on the Collaborative Topic Regression (CTR) model, a hybrid recommendation algorithm that combines a method of Collaborative Filtering (CF) and probabilistic topic modeling. The Learning Style is incorporated into the CTR to predict LO classification. The proposed model controls which classifications are more effective in the students’ learning process and which LO recommendations fit better to the student’s learning profile. Experiments were carried out in a real-world dataset collected from a Virtual Learning Environment (VLE) that was based on the inventory proposed by Felder and Soloman.

Index Terms - Hybrid Recommender System, Learning Style, Collaborative Topic Regression.