Article: Interactive teaching system based on an improved collaborative filtering algorithm and matrix decomposition Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.71 pp.56 - 79 Abstract: Providing accurate personalised resource recommendations for learners has become a key issue in improving teaching effectiveness. However, existing methods still face problems such as heavy computational burden, sparse data, and cold start when dealing with large-scale dynamic learning scenarios. Therefore, this study proposes a hybrid model integrating improved collaborative filtering, enhanced matrix decomposition, and knowledge mapping. The experimental results showed that the proposed model achieved a precision of 0.732, recall of 0.705, F1-score of 0.718, and RMSE of 0.721. For cold-start users, highly relevant recommendations increased from 28% to 64.6%. In practical teaching applications, knowledge mastery improved by 21.5%, and quiz scores increased by 12.7%. The proposed hybrid model exhibits strong robustness. It also exhibits teaching adaptability and practical value. The model improves recommendation accuracy and learning effectiveness. It provides an efficient and feasible technical solution for personalised resource recommendation in intelligent teaching systems. Inderscience Publishers - linking academia, business and industry through research

Title: Interactive teaching system based on an improved collaborative filtering algorithm and matrix decomposition

Authors: Hui Zhang; Min Lu

Addresses: School of Art and Design, Guilin University of Electronic Technology, Guilin, 541004, China ' School of Arts and Communication, Nanning College of Technology, Nanning, 541006, China

Abstract: Providing accurate personalised resource recommendations for learners has become a key issue in improving teaching effectiveness. However, existing methods still face problems such as heavy computational burden, sparse data, and cold start when dealing with large-scale dynamic learning scenarios. Therefore, this study proposes a hybrid model integrating improved collaborative filtering, enhanced matrix decomposition, and knowledge mapping. The experimental results showed that the proposed model achieved a precision of 0.732, recall of 0.705, F1-score of 0.718, and RMSE of 0.721. For cold-start users, highly relevant recommendations increased from 28% to 64.6%. In practical teaching applications, knowledge mastery improved by 21.5%, and quiz scores increased by 12.7%. The proposed hybrid model exhibits strong robustness. It also exhibits teaching adaptability and practical value. The model improves recommendation accuracy and learning effectiveness. It provides an efficient and feasible technical solution for personalised resource recommendation in intelligent teaching systems.

Keywords: collaborative filtering; CF; matrix decomposition; knowledge graph; KG; interactive teaching system; personalised recommendation.

DOI: 10.1504/IJICT.2026.154475

International Journal of Information and Communication Technology, 2026 Vol.27 No.71, pp.56 - 79

Received: 04 Jan 2026
Accepted: 02 Mar 2026
Published online: 29 Jun 2026 *

Title: Interactive teaching system based on an improved collaborative filtering algorithm and matrix decomposition

Keep up-to-date