基于学生开源社区行为的数字岗位就业预测

doi:10.3969/j.issn.1000-5641.2025.05.009

摘要/Abstract

摘要：

预测学生职业走向对于高校人才培养与企业招聘策略具有重要意义. 现有学生的就业预测大多依赖在校期间的学业数据或者其他校园行为数据, 忽略了学生开源贡献行为在数字类岗位求职过程中的作用. 为此, 基于学生的开源社区行为数据, 开展了面向数字岗位的就业预测研究. 具体而言, 文章构建了一个包含学生、代码仓库及其多种语义关系的异构信息网络来提取学生的技术特征, 并探索了两类融合大语言模型 (Large Language Model, LLM) 与图神经网络 (Graph Neural Network, GNN) 的建模策略, 大语言模型分别作为编码器 (LLM-as-Encoder) 和解释器 (LLM-as-Explainer) 来预测学生毕业后可能从事的数字岗位. 在构建的数据集上开展的大量实验表明, 所提方法在准确率和Macro-F₁上较对比方法分别提升了7.71%和9.19%. 从开源参与角度为高校就业指导提供了数据驱动的决策支持, 帮助企业精准识别技术人才, 并为学生的职业规划提供量化参考.

关键词: 开源社区行为, 学生就业预测, 异构信息网络, 图神经网络, 大语言模型

Abstract:

Accurately predicting students’ post-graduation career paths plays a vital role in talent development in higher education and in refining recruitment strategies in industry. Most existing employment prediction research relies heavily on academic or campus-related data, while overlooking the role of students’ open-source contributions in the process of securing digital-related positions. This study addresses employment prediction for digital roles by analyzing students’ behaviors in open-source communities. We construct a heterogeneous graph comprising student nodes, code repository nodes, and various semantic relationships to model students’ expertise. To enhance prediction performance, we propose two strategies that integrate large language model (LLM) with graph neural networks: LLM-as-Encoder and LLM-as-Explainer. Experiments on our curated dataset show that the proposed approach outperforms baseline methods, achieving improvements of 7.71% in accuracy and 9.19% in Macro-F₁. By leveraging open-source activity, this study supports data-driven decision-making for university career services, aids enterprises in identifying technical talent, and provides students with actionable insights for career planning.

Key words: open-source community behavior, student employment prediction, heterogeneous information network, graph neural network, large language model

中图分类号:

TP391

谢林娜, 陆雪松. 基于学生开源社区行为的数字岗位就业预测[J]. 华东师范大学学报（自然科学版）, 2025, 2025(5): 87-98.

Linna XIE, Xuesong LU. Student employment prediction for digital jobs based on behavior in open-source communities[J]. J* E* C* N* U* N* S*, 2025, 2025(5): 87-98.

图/表 10

图1

表1

图2

图3

图4

图5

图6

表2

图7

图8

参考文献 28

1	王秋霞. 2024年中国高校毕业人数统计及就业意愿、就业现状分析: 高校毕业人数再创新高, 国家加大针对毕业生职业技术培训力度 [EB/OL]. (2024-08-31)[2025-06-27]. https://www.chyxx.com/industry/1196296.html.
2	LIU R, RONG W G, OUYANG Y X, et al.. A hierarchical similarity based job recommendation service framework for university students. Frontiers of Computer Science, 2017, 11 (5): 912- 922.
3	KONG J, REN M, LU T, et al. Analysis of college students’ employment, unemployment and enrollment with self-organizing maps [M]// E-Learning and Games. Cham: Springer International Publishing, 2019: 318-321.
4	SOUMYA M D, SUGATHAN T, BIJLANI K. Improve student placement using job competency modeling and personalized feedback [C]// 2017 International Conference on Advances in Computing, Communications and Informatics (ICACCI). IEEE, 2017: 1751-1755.
5	GUO T, XIA F, ZHEN S H, et al.. Graduate employment prediction with bias. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34 (1): 670- 677.
6	YANG G Z, OUYANG Y, YE Z W, et al.. Social-path embedding-based transformer for graduation development prediction. Applied Intelligence, 2022, 52 (12): 14119- 14136.
7	UOSAKI N, MOURI K, YIN C J, et al. Seamless support for international students’ job hunting in Japan using learning log system and eBook [C]// 2018 7th International Congress on Advanced Applied Informatics (IIAI-AAI). IEEE, 2018: 374-377.
8	LIU L, SILVER D, BEMIS K.. Application-driven design: Help students understand employment and see the “big picture”. IEEE Computer Graphics and Applications, 2018, 38 (3): 90- 105.
9	Linux Foundation Research Team. The 10th annual open source jobs report [R/OL]. (2022-06-01)[2025-06-27]. https://www.linuxfoundation.org/research/the-10th-annual-open-source-jobs-report.
10	DEY T, KARNAUCH A, MOCKUS A. Replication package for representation of developer expertise in open source software [C]// 2021 IEEE/ACM 43rd International Conference on Software Engineering: Companion Proceedings (ICSE-Companion). IEEE, 2021: 995-1007.
11	DESHPANDE S, DESHMUKH S, PHADATARE A, et al.. Prediction of suitable career for students using machine learning. International Research Journal of Engineering and Technology, 2021, 8 (2): 2043- 2046.
12	REN X B, TANG J B, YIN D W, et al. A survey of large language models for graphs [C]// Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2024: 6616-6626.
13	CHEN Z K, MAO H T, LI H, et al.. Exploring the potential of large language models (LLMs) in learning on graphs. ACM SIGKDD Explorations Newsletter, 2024, 25 (2): 42- 61.
14	ZHU Y, WANG Y K, SHI H Z, et al. Efficient tuning and inference for large language models on textual graphs [C]// Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence Organization, 2024: 5734-5742.
15	HE X, BRESSON X, LAURENT T, et al. Harnessing explanations: LLM-to-LM interpreter for enhanced text-attributed graph representation learning [C]// International Conference on Learning Representations. 2024.
16	ZHANG J X, LIU J Y, LUO D S, et al. LLMExplainer: Large language model based bayesian inference for graph explanation generation [EB/OL]. (2024-07-23)[2025-06-27]. https://arxiv.org/abs/2407.15351.
17	TANG J B, YANG Y H, WEI W, et al. GraphGPT: Graph instruction tuning for large language models [C]// Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. ACM, 2024: 491-500.
18	CHEN R J, ZHAO T, JAISWAL A K, et al. LLaGA: Large language and graph assistant [C]// Proceedings of the 41st International Conference on Machine Learning. PMLR, 2024: 7809-7823.
19	LI Y H, WANG P S, LI Z X, et al. ZeroG: Investigating cross-dataset zero-shot transferability in graphs [C]// Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. ACM, 2024: 1725-1735.
20	LIU H, FENG J R, KONG L C, et al. One for all: Towards training one graph model for all classification tasks [EB/OL]. (2024-07-12)[2025-07-01]. https://arxiv.org/abs/2310.00149.
21	LIU X, WANG Y, DONG Q W, et al. Job title prediction as a dual task of expertise prediction in open source software [M]// Machine Learning and Knowledge Discovery in Databases. Applied Data Science Track. Cham: Springer Nature Switzerland, 2024: 381-396.
22	FATEMI B, HALCROW J, PEROZZI B. Talk like a graph: Encoding graphs for large language models [EB/OL]. (2023-10-06)[2025-07-01]. https://doi.org/10.48550/arXiv.2310.04560.
23	LAVALLEY M P.. Logistic regression. Circulation, 2008, 117 (18): 2395- 2399.
24	CORTES C, VAPNIK V.. Support-vector networks. Machine Learning, 1995, 20 (3): 273- 297.
25	BREIMAN L.. Random forests. Machine Learning, 2001, 45 (1): 5- 32.
26	HAMILTON W L, YING Z, LESKOVEC J. Inductive representation learning on large graphs [C]// Proceedings of the 30th Annual Conference on Neural Information Processing Systems (NeurIPS). NeurIPS, 2017: 1025–1035.
27	KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks [EB/OL]. (2017-02-22)[2025-07-01]. https://arxiv.org/pdf/1609.02907.
28	VELIČKOVIĆ P, CUCURULL G, CASANOVA A, et al. Graph attention networks [EB/OL]. (2018-02-04)[2025-07-01]. https://doi.org/10.48550/arXiv.1710.10903.

Student		Repository
Student ID	Education	Repo ID	API
0	"school": "Arizona State University", "start_date": "2011", "end_date": "2015", "degrees": ["bachelors", "masters"], "majors": ["Science in Software Engineering"]	0	javax.servlet.http.HttpServlet, org.springframework.boot, org.springframework.web, $\cdots $
		1	javax.crypto.Cipher, javax.crypto.KeyGenerator, java.nio.file.Paths, java.util.Base64, org.bouncycastle.jce.provider.BouncyCastleProvide, $\cdots $
		2	sklearn.model_selection.train_test_split, sklearn.linear_model.LogisticRegression, sklearn.metrics.accuracy_score, $\cdots $

组别	方法	Accuracy	Macro-F₁	Micro-F₁
ML	SVM	20.81	15.38	20.81
	LR	22.43	16.24	22.43
	RF	24.46	17.65	24.46
GNN	GAT	28.37	19.87	23.69
	GraphSAGE	28.98	20.21	24.35
	GCN	29.86	20.54	24.89
LLM	Mistral-7B	34.75	27.84	34.75
	Qwen-7B	35.62	28.52	35.62
	DeepSeek-R1-Distill-Qwen-7B	36.41	29.04	36.41
LLM + GNN	LLM-as-Encoder	42.33	35.57	37.94
LLM + GNN	LLM-as-Explainer	44.12	38.23	40.11

[1]	许立君, 杨丽, 黄子祎. 大语言模型与开源生态协同的人工智能课程创新探索与研究[J]. 华东师范大学学报（自然科学版）, 2025, 2025(5): 66-75.
[2]	庞瑞洋, 陆雪松. 基于智能体的可交互数据结构和算法可视化实现[J]. 华东师范大学学报（自然科学版）, 2025, 2025(5): 32-42.
[3]	何德鑫, 韩凡宇, 王伟. 大语言模型在开源项目主题标注中的应用与评估研究[J]. 华东师范大学学报（自然科学版）, 2025, 2025(5): 14-24.
[4]	范厚龙, 房爱莲, 林欣. 文本信息与图结构信息相融合的知识图谱补全[J]. 华东师范大学学报（自然科学版）, 2025, 2025(1): 111-123.
[5]	陈郅睿, 陆雪松. 基于开源代码大语言模型提示的学生代码修复[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 93-103.
[6]	寇思佳, 闫凤云, 马晶. 国内大语言模型在学科知识图谱自动标注上的应用——以道德与法治和数学学科为例[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 81-92.
[7]	刘佳, 孙新, 张宇晴. 知识图谱与大语言模型协同的教育资源内容审查[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 57-69.
[8]	孔超, 陈家会, 孟丹, 刁华彬, 王维, 张丽平, 刘涛. 面向MOOCs的个性化知识概念推荐[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 32-44.