Article: Vocational education course recommendation based on neighbours under the construction of knowledge graph Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.40 pp.1 - 17 Abstract: Focusing on issues of incomplete user information and inaccurate recommendations in current vocational education course recommendation methods, this paper first constructs a knowledge graph (KG) for vocational education. On this basis, a novel negative sampling approach is employed to enhance the KG representation model TransH (EOTransH), and different weights are given to negative samples assigned different scores contribute to full model training. Then, a bipartite graph of courses and users is constructed, and KG embedding is built through the joint user entity neighbourhood information. Furthermore, higher-order connectivity information between users and courses is mined through attention-based propagation. Finally, an attention network is built in the output prediction layer to explore user preference features. Experimental outcome on the MOOCCourse and MOOCCube datasets indicate that the proposed approach improves F1 by at least 2.05%, 4.09%, effectively solving the problem of inaccurate recommendations. Inderscience Publishers - linking academia, business and industry through research

Title: Vocational education course recommendation based on neighbours under the construction of knowledge graph

Authors: Hantian Wang

Addresses: Engineering Technology Training Center, Nanjing University of Industry Technology, Nanjing, 210023, China

Abstract: Focusing on issues of incomplete user information and inaccurate recommendations in current vocational education course recommendation methods, this paper first constructs a knowledge graph (KG) for vocational education. On this basis, a novel negative sampling approach is employed to enhance the KG representation model TransH (EOTransH), and different weights are given to negative samples assigned different scores contribute to full model training. Then, a bipartite graph of courses and users is constructed, and KG embedding is built through the joint user entity neighbourhood information. Furthermore, higher-order connectivity information between users and courses is mined through attention-based propagation. Finally, an attention network is built in the output prediction layer to explore user preference features. Experimental outcome on the MOOCCourse and MOOCCube datasets indicate that the proposed approach improves F1 by at least 2.05%, 4.09%, effectively solving the problem of inaccurate recommendations.

Keywords: vocational education course recommendation; knowledge graph; nearest neighbour method; TransH model; attention network.

DOI: 10.1504/IJICT.2025.149811

International Journal of Information and Communication Technology, 2025 Vol.26 No.40, pp.1 - 17

Received: 26 Jul 2025
Accepted: 18 Sep 2025
Published online: 13 Nov 2025 *

Title: Vocational education course recommendation based on neighbours under the construction of knowledge graph

Keep up-to-date