Competency Analysis System for Higher Vocational Art and Design Teachers Based on Collaborative Filtering with K-Nearest Neighbor

In the recent days, the sector of higher vocational education in art and design needs highly skilled teachers with specific competencies to prepare students efficiently. Previous researches have suggested various effective methods but still there are challenges such as recognizing and standardizing key performance indicators that align with higher vocational teaching demands while enabling adaptive feedback for teachers' continuous improvement. This study aims to design and construct a competency analysis system for higher vocational art and design teachers using deep learning to effectively capture complex teaching abilities through large-scale data analysis. Initially the data is collected from Teaching and Learning International Survey (TALIS). Then, the collected data is preprocessed using data standardization and tokenization. The processed data is further fed into feature extraction model namely, Robustly optimized Bidirectional Encoder Representations from Transformers (RoBERT) since it can effective extracts the optimal attributes and assist for analysis. In the end, Collaborative filtering with K-Nearest Neighbor (KNN) is utilized for recommendation model and analyzed using the evaluation metrics such as accuracy, recall, precision and fl-score. The proposed model is compared with various conventional methods namely BERT and resulted in higher accuracy of 98.67%.