Article: Knowledge graph-based adaptive recommendation model for training courses Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.3 pp.53 - 69 Abstract: The contradiction between course resource overload and learners' personalised needs in online education platforms is becoming increasingly prominent. Addressing the common issues of weak interpretability and poor dynamic adaptability in existing recommendation methods, this paper proposes a knowledge graph-based adaptive course recommendation model. By constructing a hierarchical knowledge graph to precisely represent the course knowledge system and integrating deep knowledge tracking with reinforcement learning techniques, the model dynamically perceives learners' knowledge states and evolving interests, enabling real-time adjustment of recommendation paths. Experiments on the publicly available china university massive open online course dataset demonstrate that compared to mainstream baseline models, our model achieves up to 8.7% higher performance on key metrics such as normalised discounted cumulative gain@10 and hit rate @10. This validates its effectiveness and superiority in delivering precise, explainable personalised recommendations. Inderscience Publishers - linking academia, business and industry through research

Title: Knowledge graph-based adaptive recommendation model for training courses

Authors: Zhenhong Li

Addresses: Department of Basic Courses, Shandong Institute of Commerce and Technology, Jinan, 250103, China

Abstract: The contradiction between course resource overload and learners' personalised needs in online education platforms is becoming increasingly prominent. Addressing the common issues of weak interpretability and poor dynamic adaptability in existing recommendation methods, this paper proposes a knowledge graph-based adaptive course recommendation model. By constructing a hierarchical knowledge graph to precisely represent the course knowledge system and integrating deep knowledge tracking with reinforcement learning techniques, the model dynamically perceives learners' knowledge states and evolving interests, enabling real-time adjustment of recommendation paths. Experiments on the publicly available china university massive open online course dataset demonstrate that compared to mainstream baseline models, our model achieves up to 8.7% higher performance on key metrics such as normalised discounted cumulative gain@10 and hit rate @10. This validates its effectiveness and superiority in delivering precise, explainable personalised recommendations.

Keywords: knowledge graph; adaptive recommendation; massive open online courses; MOOCs; personalised learning.

DOI: 10.1504/IJICT.2026.151492

International Journal of Information and Communication Technology, 2026 Vol.27 No.3, pp.53 - 69

Received: 20 Oct 2025
Accepted: 16 Nov 2025
Published online: 02 Feb 2026 *

Title: Knowledge graph-based adaptive recommendation model for training courses

Keep up-to-date