Person-job Matching Model Enhanced by Dynamic Job Knowledge Graph

DOI: https://doi.org/10.62517/jiem.202603311

Author(s)

Zongxing Zhang1, Xiyao Zheng1, Yixin Yan1, Jinjia Lu1, Zhen Hao2

Affiliation(s)

1School of Artificial Intelligence and Big Data, Henan University of Technology, Zhengzhou, China 2IFLYTEK Co., Ltd., Hefei, China

Abstract

The persistent two-way mismatch between enterprise recruitment and job seekers' career choices in the labor market, compounded by the fact that technological iteration generally outpaces the renewal cycle of talent cultivation, has placed mounting pressure on precise person-job matching. In the practical operation of recruitment, this contradiction manifests in three key issues: the absence of standardized quality control mechanisms in the fusion of multi-source heterogeneous data, the lack of reliable factual boundary constraints on content generated by large language models, and the difficulty of quantitatively tracking the dynamic evolution of job competency requirements. To address these three issues, this paper proposes a knowledge-graph-grounded large language model enhancement method to alleviate multiple complex problems in person-job matching. The method comprises four components. In terms of data governance, a five-factor quality assessment framework and a multi-field composite similarity algorithm are employed to transform data of various formats into a structured knowledge graph. In terms of temporal evolution, snapshot comparison and trend scoring are utilized to track changes in job competency requirements. In terms of hallucination prevention, the knowledge graph serves as the factual basis for a series of operations—including input filtering, output verification, graph reverse lookup, and credibility scoring—thereby forming a multi-level iterative verification pipeline. In terms of matching, a seven-dimensional weighted scoring system is designed by integrating Holland's vocational interest theory and the Iceberg competency model. Validation experiments were conducted on a standard test set constructed from 100 annotated job descriptions (JDs) and 400 candidate samples, achieving an F1 score of 95.0% for JD parsing, 96.8% for resume information extraction, and a precision of 92.8% for person-job matching. The experimental results demonstrate that the adopted hybrid entity extraction scheme, which combines domain dictionaries with LLM assistance, achieves an F1 score of 96.8% on the skill extraction task.

Keywords

Knowledge Graph; Multi-Source Heterogeneous Data Fusion; Hallucination Suppression; Dynamic Evolution Analysis

References

[1] Zhang Dandan, Yu Hang, Li Lixing, et al. The Measurement of AI Exposure and Its Impact on Labor Demand in China: Evidence from Large Language Models. Management World, 2025, 41(7): 59-75. [2] Chen Lin, Gao Yuepeng, Yu Linhui. How Is AI Shaping the Labor Demands of Enterprises? Evidence from Big Data Analysis of Recruitment Platforms. Management World, 2024, 40(6): 74-93. [3] Yao Jiaquan, Zhang Kunpeng, Guo Lipeng, et al. How Does Artificial Intelligence Improve Firm Productivity? Based on the Perspective of Labor Skill Structure Adjustment. Management World, 2024, 40(2): 101-116, 117-122, 133. [4] Wang Yongqin, Dong Wen. How the Rise of Robots Has Affected China's Labor Market: Evidence from China's Listed Manufacturing Firms. Economic Research Journal, 2020, 55(10): 159-175. [5] Tian Ge, Zhang Xun. Digital Economy, Non-agricultural Employment, and Division of Labor. Management World, 2022, 38(5): 72-84, 311. [6] Liu Qiao, Li Yang, Duan Hong, et al. Knowledge Graph Construction Techniques. Journal of Computer Research and Development, 2016, 53(3): 582-600. [7] Xu Zenglin, Sheng Yongpan, He Lirong, Wang Yafang. Review on Knowledge Graph Techniques. Journal of University of Electronic Science and Technology of China, 2016, 45(4): 589-606. [8] Shen Kejie, Huang Huanting, Hua Bolin. Constructing Knowledge Graph with Public Resumes. Data Analysis and Knowledge Discovery, 2021, 5(7): 81-90. [9] Qin Chuan, Zhu Hengshu, Zhuang Fuzhen, et al. A Survey on Knowledge Graph-Based Recommender Systems. Scientia Sinica Informationis, 2020, 50(7): 937-956. [10] Xu Mengqi, Xiong Xi, Li Binyong, et al. Construction of Person-Post Matching Recommended System Based on Knowledge Graph. Application Research of Computers, 2022, 39(1): 194-198. [11] Wang Yuhan, Chen Ziyang, Zhao Xiang, et al. Research Progress and Trend of Temporal Knowledge Graph Representation and Reasoning. Journal of Software, 2024, 35(8): 3923-3951. [12] Shen Yinghan, Jiang Xuhui, Wang Yuanzhuo, et al. A Survey of Temporal Knowledge Graph Reasoning. Chinese Journal of Computers, 2023, 46(6): 1272-1301. [13] Tian Yonglin, Wang Yutong, Wang Xingxia, et al. From Retrieval-augmented Generation to SAGE: The State of the Art and Prospects. Acta Automatica Sinica, 2025, 51(6): 1145-1169. [14] Huang Bo, Wu Shen'ao, Wang Wenguang, et al. KG-LLM-MCom: A Survey on Integration of Knowledge Graph and Large Language Model. Journal of Wuhan University (Natural Science Edition), 2024, 70(4): 397-412. [15] Liu Zeyuan, Wang Pengjiang, Song Xiaobin, et al. Survey of Hallucination in Large Language Models. Journal of Software, 2025, 36(3): 1152-1185. [16] Fang Yanran, Xie Xiaoyun, Shi Junqi. Human Resource Management Research in the Era of Generative Artificial Intelligence: A Work-Flow Analysis Perspective. Bulletin of National Natural Science Foundation of China, 2024, 38(5): 820-830. [17] Zhao Yixuan, Zhao Shuming, Luan Jiarui. Human Resource Management Based on Artificial Intelligence: A Conceptional Model and Future Research. Nanjing Journal of Social Sciences, 2020, (2): 36-43. [18] Yang Yanwu, Zhai Panyu, Jin Ying. Entity Recognition and Talent Profiles in Digital Industry Job Based on BERT and BiGRU. Journal of Wuhan University (Natural Science Edition), 2025, 71(5): 667-679. [19] Meng Xiaofeng, Du Zhijuan. Research on the Big Data Fusion: Issues and Challenges. Journal of Computer Research and Development, 2016, 53(2): 231-246. [20] Wu Jianbin, Shen Juanfeng. Analysis of College Students' Vocational Personality Based on Holland's Personality Type Theory. Education and Vocation, 2014, (8): 87-89. [21] Xu Feng. The Construction of HR Performance Management System: Based on the Competency Model. Enterprise Economy, 2012, 31(1): 68-71. [22] Zhang Rujia, Dai Lu, Wang Bang, et al. Review of Recent Research Progress in Chinese Named Entity Recognition Based on Deep Learning. Journal of Chinese Information Processing, 2022, 36(6): 20-35.