Design and Application of Online Teaching Platform for Vocational English using Robust Bidirectional Encoder Representations from Transformers with Collaborative Filtering

In recent years, blended teaching has emerged as an educational approach that combines the advancements of both offline and online teaching. However, this type of teachings has their individual advantages but also has limitations of time and space. To overcome these problems, in this research, Robust Bidirectional Encoder Representations from Transformers with Neural Collaborative Filtering (ROBERTA-NCF) is proposed for recommending Online teaching platform for vocational English. Initially dataset is collected from an online blended teaching platform which contains details of English related topics. Further, data is pre-processed by utilizing cleaning, tokenization and lemmatization to identify and correcting errors, to divide data into tokens and to braking words into their roots. After that, data is fed into RoBERTa for feature extraction by utilizing transforming based architecture. Finally, CF is employed for building effective recommendation system and optimize the current English teaching mode, thus improving students English learning level. The proposed ROBERTA-NCF achieved improved results in terms of Match Rate (85.00%) and Test Pass Rate (91.50%) when compared to existing Collaborative Filtering Algorithm (CFA).