BTCBMA Online Education Course Recommendation Algorithm Based on Learners' Learning Quality

Abstract

To address the problems of existing online education curriculum recommendation methods such as low recommendation accuracy, an online education course recommendation algorithm (BTCBMA) considering learner learning quality is proposed. Firstly, the BERT model is combined with the TextCNN model to implement the preliminary extraction of text features. Secondly, the convolution neural networks and BiLSTM networks are used to capture deep features and temporal features in data. Finally, a multi-head attention mechanism is used to extract key information from learner interaction sequences, review texts, and curriculum multiple attributes. Experiments demonstrate that the accuracy, precision, recall, and F1 values of the proposed online course recommendation method in the MOOC dataset are 0.224, 0.241, 0.237, and 0.239, respectively, while in the CN dataset are 0.217, 0.239, 0.227, and 0.233, respectively, and the performance of the proposed method in online education course recommendation is significantly superior to the compared methods. For learners in online learning systems, the proposed method can effectively recommend high-quality courses, which is of great significance for improving the learning quality and learning efficiency of learners.