Article: Risk perception and traceability in higher education quality management using dynamic knowledge graphs and link prediction Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.98 pp.76 - 93 Abstract: Higher education management risks are complex and dynamic, making accurate root-cause tracing difficult. This paper proposes the academic risk intelligence and traceability agent (ARITA) model, integrating dynamic knowledge graphs and link-prediction to create a closed-loop framework for risk perception and traceability. Using over 500,000 multi-source data records, we constructed a temporal knowledge graph with over 100,000 entities. ARITA utilises dynamic graph construction, attention-enhanced risk perception, and reinforcement learning-based intelligent agent traceability. Experiments demonstrate ARITA's superiority over the AttBiTi model, achieving an area under the curve (AUC) of 0.952 and an F1-score of 0.891 in risk link prediction. Furthermore, it attains an 85% success rate in tracing risk root causes, effectively restoring conduction paths with an average length of 3.5 hops. This research offers an intelligent, data-driven decision-making paradigm for higher education quality management. Inderscience Publishers - linking academia, business and industry through research

Title: Risk perception and traceability in higher education quality management using dynamic knowledge graphs and link prediction

Authors: Hua Zhong

Addresses: Gannan Medical University, Ganzhou, 341000, China

Abstract: Higher education management risks are complex and dynamic, making accurate root-cause tracing difficult. This paper proposes the academic risk intelligence and traceability agent (ARITA) model, integrating dynamic knowledge graphs and link-prediction to create a closed-loop framework for risk perception and traceability. Using over 500,000 multi-source data records, we constructed a temporal knowledge graph with over 100,000 entities. ARITA utilises dynamic graph construction, attention-enhanced risk perception, and reinforcement learning-based intelligent agent traceability. Experiments demonstrate ARITA's superiority over the AttBiTi model, achieving an area under the curve (AUC) of 0.952 and an F1-score of 0.891 in risk link prediction. Furthermore, it attains an 85% success rate in tracing risk root causes, effectively restoring conduction paths with an average length of 3.5 hops. This research offers an intelligent, data-driven decision-making paradigm for higher education quality management.

Keywords: dynamic knowledge graph; link prediction; risk traceability; ARITA model; higher education management.

DOI: 10.1504/IJICT.2026.156256

International Journal of Information and Communication Technology, 2026 Vol.27 No.98, pp.76 - 93

Received: 03 Apr 2026
Accepted: 27 Apr 2026
Published online: 08 Sep 2026 *

Title: Risk perception and traceability in higher education quality management using dynamic knowledge graphs and link prediction

Keep up-to-date