Article: Joint modelling and governance of knowledge graphs and reasoning rules for vocational skill assessment Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.47 pp.78 - 99 Abstract: Vocational skill assessment needs decisions that are accurate, explainable, and stable when rubrics change. To reduce hidden contradictions and cohort drift in evidence-driven assessment, this paper proposes a joint management approach that integrates a knowledge graph with rule-checked reasoning and distribution monitoring. First, indicator dependencies and evidence links are organised into a graph to keep decisions traceable. Then, explicit rules are validated and applied together with model outputs to prevent inconsistent judgements. Finally, evidence embeddings are analysed with manifold and density diagnostics to reveal sparse regions and guide targeted governance tests. Experiments on five cohorts and large-scale interaction logs show that decision accuracy increases by 3.4% points, rule-consistency improves from 88.1% to 96.7%, and uncovered rule paths decrease by 41% compared with strong baselines. The approach supports maintainable, audit-ready assessment at scale. Inderscience Publishers - linking academia, business and industry through research

Title: Joint modelling and governance of knowledge graphs and reasoning rules for vocational skill assessment

Authors: Yi Zhang; Menglu Wang; Zhongqiu Guan; Weiping Zhang

Addresses: Organization and Publicity Office, Qinhuangdao Vocational and Technical College, Hebei 066100, China ' Department of Mechanical and Electrical Engineering, Qinhuangdao Vocational and Technical College, Hebei 066100, China ' Department of Economics, Qinhuangdao Vocational and Technical College, Hebei 066100, China ' Department of Mechanical and Electrical Engineering, Qinhuangdao Vocational and Technical College, Hebei 066100, China

Abstract: Vocational skill assessment needs decisions that are accurate, explainable, and stable when rubrics change. To reduce hidden contradictions and cohort drift in evidence-driven assessment, this paper proposes a joint management approach that integrates a knowledge graph with rule-checked reasoning and distribution monitoring. First, indicator dependencies and evidence links are organised into a graph to keep decisions traceable. Then, explicit rules are validated and applied together with model outputs to prevent inconsistent judgements. Finally, evidence embeddings are analysed with manifold and density diagnostics to reveal sparse regions and guide targeted governance tests. Experiments on five cohorts and large-scale interaction logs show that decision accuracy increases by 3.4% points, rule-consistency improves from 88.1% to 96.7%, and uncovered rule paths decrease by 41% compared with strong baselines. The approach supports maintainable, audit-ready assessment at scale.

Keywords: vocational skill assessment; knowledge graph; rule-based reasoning; evidence governance; embedding manifold; robustness monitoring.

DOI: 10.1504/IJICT.2026.153525

International Journal of Information and Communication Technology, 2026 Vol.27 No.47, pp.78 - 99

Received: 28 Jan 2026
Accepted: 28 Feb 2026
Published online: 12 May 2026 *

Title: Joint modelling and governance of knowledge graphs and reasoning rules for vocational skill assessment

Keep up-to-date