Abstract
Modern, efficient techniques of identifying and regulating digital Educational Resources (ER) are urgently needed due to the rapid development of this area of study. Due to the level of detail and emphasis on information, conventional Text Classification (TC) techniques frequently experience problems when working with ER. To overcome these challenges, this study proposes a novel approach termed BERT-CNN, which integrates the effectiveness of BERT's knowledge of context with CNN's pattern identification skills in order to attain Deep Learning (DL). The objective of the hybrid approach is to improve educational Text Labelling (TL) and TC to be more precise, successful, and robust. The BERT-CNN model is superior to conventional approaches in various measures, including accuracy, precision, recall, F1 score, and AUC-ROC, based on thorough testing and comparison. Providing essential insights into the possibilities of integrating BERT and CNN for educational purposes, this research covers detailed methods from data collection to model development, testing, and the implementation of modern TC approaches.
Published Version (Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have