Evidence-Constrained Construction and Evolution Analysis of Job Competency Graphs

DOI: https://doi.org/10.62517/jes.202602401

Author(s)

Ge Tian1, Anhong Chen1, Hanxiao Cao1, Zhen Hao2,*

Affiliation(s)

1Henan University of Technology, Zhengzhou, Henan, China 2iFLYTEK Co., Ltd., Hefei, Anhui, China *Corresponding Author.

Abstract

To address hallucinated facts, miss-ing evidence chains, unauditable versions, and unquantified uncertainty in job competency graphs, this study develops an integrated framework using 15,709 job postings from Zhaopin, ByteDance Careers, and BOSS Zhipin. The framework combines data denoising, evidence binding, review and release, versioned storage, and trend forecasting. It uses a three-component weighted similarity measure to control textual noise, nine-dimensional evidence metadata and a draft-review-release-discard state machine to govern graph publication, and an ensemble of weighted linear trend regression and exponentially weighted moving average for skill-demand forecasting. Experiments produced a versioned graph with 15,766 entities and 29,667 relations. Compared with the baseline, duplicate-detection F1 increased by 11.4%; 78% of published relations were traceable to source evidence; and the rate of erroneous forecast delivery decreased by 62% in the three-month minimum-sample setting. The framework supports traceability, version auditing, and interpretable trend analysis for talent inventory and vocational education planning.

Keywords

Job Competency Graph; Large Language Model Hallucination; Evidence Constraints; Version Management; Uncertainty-Aware Time-Series Forecasting

References

[1] Zhang Wen, Lai Kee-Hung, Gong Qiguo. The Future of the Labor Force: Higher Cognition and More Skills. Humanities and Social Sciences Communications, 2024, 11(1):Article479.DOI:10.1057/s41599-024-02962-1. [2] Vrolijk J, Poslavsky V, Bijl T, et al. Ontology Learning for ESCO: Leveraging LLMs to Navigate Labor Dynamics. Joint Proceedings of the 2nd Workshop on Knowledge Base Construction from Pre-Trained Language Models and the 3rd Challenge on Language Models for Knowledge Base Construction (KBC-LM 2024 and LM-KBC 2024), Baltimore, USA. CEUR Workshop Proceedings, 2024, 3853: 1-16. [3] Karakeben K, Sorotos H E, Milchevskaya A, et al. Rapid Graph Generation from Job Descriptions: Combining Taxonomies and LLMs//The Semantic Web: ESWC 2024 Satellite Events. Lecture Notes in Computer Science, Vol. 15344. Cham: Springer, 2025: 143-146.DOI:10.1007/978-3-031-78952-6_15. [4] Wang Y, Wang M, Manzoor M A, et al. Factuality of Large Language Models: A Survey//Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024). Miami, USA: Association for Computational Linguistics, 2024: 19519-19529.DOI:10.18653/v1/2024.emnlp-main.1088. [5] Nguyen K, Zhang M, Montariol S, et al. Rethinking Skill Extraction in the Job Market Domain Using Large Language Models//Proceedings of the First Workshop on Natural Language Processing for Human Resources (NLP4HR 2024). St. Julian's, Malta: Association for Computational Linguistics, 2024: 27-42.DOI:10.18653/v1/2024.nlp4hr-1.3. [6] Autor David, Levy Frank, Murnane Richard. The Skill Content of Recent Technological Change: An Empirical Exploration. The Quarterly Journal of Economics, 2003, 118(4):1279-1333.DOI:10.1162/003355303322552801. [7] Cuconasu F, Trappolini G, Siciliano F, et al. The Power of Noise: Redefining Retrieval for RAG Systems//Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2024). Washington, D.C., USA: Association for Computing Machinery, 2024: 719-729.DOI:10.1145/3626772.3657834. [8] Acemoglu D, Autor D. Skills, Tasks and Technologies: Implications for Employment and Earnings//Ashenfelter O, Card D, eds. Handbook of Labor Economics. Vol. 4B. Amsterdam: Elsevier, 2011: 1043-1171.DOI:10.1016/S0169-7218(11)02410-5. [9] Sood S, Loguinov D. Probabilistic Near-Duplicate Detection Using Simhash//Proceedings of the 20th ACM International Conference on Information and Knowledge Management (CIKM '11). Glasgow, UK: Association for Computing Machinery, 2011:1117-1126.DOI:10.1145/2063576.2063737. [10] Devlin J, Chang M W, Lee K, et al. BERT: Pre-Training of Deep Bidirectional Transformers for Language Understanding//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). Minneapolis, USA: Association for Computational Linguistics, 2019: 4171-4186.DOI:10.18653/v1/N19-1423. [11] Hyndman R J, Athanasopoulos G. Forecasting: Principles and Practice. 3rd ed. Melbourne: OTexts, 2021. [12] Dong X L, Gabrilovich E, Heitz G, et al. Knowledge Vault: A Web-Scale Approach to Probabilistic Knowledge Fusion//Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD '14). New York, USA: Association for Computing Machinery, 2014: 601-610.DOI:10.1145/2623330.2623623. [13] De-Arteaga M, Romanov A, Wallach H, et al. Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Setting//Proceedings of the Conference on Fairness, Accountability, and Transparency (FAT* '19). Atlanta, USA: Association for Computing Machinery, 2019:120-128.DOI:10.1145/3287560.3287572.