Article: Intelligent agent reinforcement learning integrated with knowledge graphs for optimising personalised learning paths Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.89 pp.63 - 86 Abstract: In online learning scenarios, the complex dependencies between knowledge concepts are often simplified, and the dynamic changes in learners' cognitive states are difficult to capture. As a result, personalised learning path recommendations generally lack the capacity to achieve global optimisation. To address this, the paper proposes a knowledge graph-enhanced reinforcement learning approach for personalised learning path optimisation. This method constructs a heterogeneous knowledge graph and introduces a graph attention network (GAT) to enhance concept representation, then employs an actor-critic agent to learn optimal decision strategies under composite reward guidance. On the MOOCCubeX dataset, the method improves Precision@5 from 0.504 to 0.547 and normalised discounted cumulative gain@5 from 0.489 to 0.535; on the MOOPer dataset, Precision@5 improves from 0.462 to 0.508, and normalised discounted cumulative gain@5 improves from 0.441 to 0.488. Inderscience Publishers - linking academia, business and industry through research

Title: Intelligent agent reinforcement learning integrated with knowledge graphs for optimising personalised learning paths

Authors: Xiaoqin Ren; Xiaoying Zhang; Wei Zhang

Addresses: Department of Health Management, Shanxi Health Vocational College, Jinzhong, 030600, China ' Department of Health Management, Shanxi Health Vocational College, Jinzhong, 030600, China; Shanxi Province Traditional Chinese Medicine School, Jinzhong, 030600, China ' Department of Health Management, Shanxi Health Vocational College, Jinzhong, 030600, China

Abstract: In online learning scenarios, the complex dependencies between knowledge concepts are often simplified, and the dynamic changes in learners' cognitive states are difficult to capture. As a result, personalised learning path recommendations generally lack the capacity to achieve global optimisation. To address this, the paper proposes a knowledge graph-enhanced reinforcement learning approach for personalised learning path optimisation. This method constructs a heterogeneous knowledge graph and introduces a graph attention network (GAT) to enhance concept representation, then employs an actor-critic agent to learn optimal decision strategies under composite reward guidance. On the MOOCCubeX dataset, the method improves Precision@5 from 0.504 to 0.547 and normalised discounted cumulative gain@5 from 0.489 to 0.535; on the MOOPer dataset, Precision@5 improves from 0.462 to 0.508, and normalised discounted cumulative gain@5 improves from 0.441 to 0.488.

Keywords: knowledge graph; reinforcement learning; personalised learning paths; agents.

DOI: 10.1504/IJICT.2026.155692

International Journal of Information and Communication Technology, 2026 Vol.27 No.89, pp.63 - 86

Received: 10 Apr 2026
Accepted: 01 Jun 2026
Published online: 10 Aug 2026 *

Title: Intelligent agent reinforcement learning integrated with knowledge graphs for optimising personalised learning paths

Keep up-to-date