Research and Innovative Application of Multi-Agent Collaborative Architecture in Education
DOI:
https://doi.org/10.54097/jpb8jw12Keywords:
Multi-agent collaboration; knowledge graph; retrieval-augmented generation; prompt engineering; workflow engine.Abstract
To address issues such as lack of dynamic adaptability and weak teaching standardization in generative AI education applications, this paper proposes an educational multi-agent collaborative system architecture. The system achieves closed-loop management of the entire teaching process through the collaborative operation of four specialized intelligent agents: "teaching, learning, administering, and evaluating". Key technological breakthroughs include knowledge graph using Neo4j-based cognitive graph, RAG-enhanced retrieval to boost semantic-level knowledge matchmaking accuracy and intelligent workflow engine to automate concatenation of teaching quests. It promotes the transformation of the education paradigm from static knowledge delivery to dynamic cognitive construction, and evolves from educator-centered to student-centered. In the future, this system will optimize the scheduling algorithm and retrieval performance of the agents and deepen the fusion path with traditional education models.
Downloads
References
[1] Liu Ming, Yang Min, Wu Zhongming, et al. The Development, Application Status and Future Prospects of Intelligent Agents for Large Educational Models [J]. Modern Educational Technology,2024, (11):5-14.
[2] Lu, Y., Yu, J. L., & Chen, P. H. (2024). Research on the construction and application of teaching intelligent agents based on large models. China Educational Technology, (7), 99-108.
[3] Liu, S. Q., Liu, Z., Duan, H. M., et al. (2025). Research on the application of large model-driven educational multi-agent systems: Technical architecture, development status, practical paths and future prospects. Journal of Distance Education, 43(01), 33-45. https://doi.org/10.15881/j.cnki.cn33-1304/g4.2025.01.004
[4] arXiv:2305.14687 [cs.IT] https://doi.org/10.48550/arXiv.2305.14687
[5] Lu, Y., Yu, J. L., & Chen, P. H. (2024). Research on the construction and application of teaching intelligent agents based on large models. China Educational Technology, (07), 99-108.
[6] Huang, C. Q., Zhong, Y. H., Wang, X. Z., Han, Z. M., & Wei, T. Q. (2025). From single agent to multi-agent: Design and empirical research on incentive-based learning activities supported by large model intelligent agents. Journal of East China Normal University (Educational Sciences Edition), 43(5), 44-56.
[7] Ba, Z. Z., Zhang, H., Xie, Z. H., et al. (2025). A review of automated prompt engineering technology for large language models. Journal of Computer Science and Exploration, (Online in Press), 1-28. Retrieved July 20, 2025, from http://kns.cnki.net/kcms/detail/11.5602.TP.20250701.1702.004.html
[8] Hao, J., Niu, H. W., Du, J., et al. (2025). Typical applications of knowledge graphs and large models in education and teaching. China Modern Educational Equipment, (11), 1-5. https://doi.org/10.13492/j.cnki.cmee.2025.11.003
[9] Gao, L. P., Cai, W. J., Zheng, W., et al. (2025). Application and practice of intelligent tutoring systems based on large models and knowledge graphs in blended teaching of software testing courses. Software Guide, (Online in Press), 1-8. Retrieved August 24, 2025, from https://link.cnki.net/urlid/42.1671.TP.20250814.1517.020
[10] Huang, H. W., Du, J., Lu, Y. T., et al. (2025). Application of semantic feature space model in intelligent question answering based on RAG. Software Guide, (Online in Press), 1-10.Retrieved July 25, 2025, from http://kns.cnki.net/kcms/detail/42.1671.tp.20250529.1651.032.html
[11] Zheng, Y. F., Zhao, Y. N., Huang, J. Y., et al. (2025). Educational intelligent agents: Research status and development trends. Modern Distance Education Research, 37(04), 3-13+59.
[12] Wan, P., & Gu, X. Q. (2024). From others’ interpretation to heteronomous identity: Practical transformation of teacher learning empowered by multi-agents. Modern Distance Education, (06), 16-24. https://doi.org/10.13927/j.cnki.yuan.20241213.001
[13] Qin, J. H., Ma, Q. C., Li, M., Zhang, C., Fu, W. M., Liu, Q. C., & Zheng, W. X. (2025). A review of multi-agent collaboration research progress: A cross-perspective of game theory and control. Acta Automatica Sinica, 51(3), 489-509. https://doi.org/10.16383/j.aas.c240508
[14] Zhang, P. L., Ma, Y. X., Li, H., et al. (2025). Intent recognition framework for multi-agent networks driven by large language models. Journal of Chinese Computer Systems, (Online in Press), 1-10. Retrieved August 1, 2025, from https://doi.org/10.20009/j.cnki.21-1106/TP.2025-0189
[15] Mei, Y. H., Wang, L. L., Wang, P. F., et al. (2024). Database knowledge question answering system based on multimodality and retrieval-augmented generation. Computer Education, (12), 232-237. https://doi.org/10.16512/j.cnki.jsjjy.2024.12.023
[16] Dan Wu, Xin Jiang, Shaobo Liang, Fei Tang, Chao Qiu, Chi Yu, Peihui Yan. Wuhan University Pioneers the “AI +” Professional Knowledge Graph Spanning the Teaching–Learning–Management–Evaluation Chain. Frontiers of Digital Education, 2025, 2(1): 3 https://doi.org/10.1007/s44366-025-0042-2
[17] Jin, J. D., Huang, Z., Hu, Z. Y., et al. (2024). Advanced phishing email detection method based on agent workflow. Journal on Communications, 45(S2), 59-68.
[18] Meng, L. Q., & Huang, Q. H. (2025). Analysis on the reform and innovation of ideological and political courses in colleges and universities empowered by knowledge graphs. Journal of Wuhan University of Technology (Social Sciences Edition), (Online in Press), 1-7. Retrieved August 14, 2025, from https://link.cnki.net/urlid/42.1660.C.20250808.0913.002
[19] Guo, Q. Z. (2022). Current situation, key points and future of virtual digital human development. News and Writing, (7), 56-64.
[20] Wu, Q., Zhou, C., Wu, T. Y., et al. (2025). Research on countermeasures for improving rural teachers’ digital literacy empowered by educational virtual digital humans. Modern Information Technology, 9(12), 141-146+151. https://doi.org/10.19850/j.cnki.2096-4706.2025.12.027
[21] Li, B. X., Lu, X., Zhang, X., et al. (2025). Hallucination detection method for retrieval-augmented private questions answering large models. Software Guide, (Online in Press), 1-8. Retrieved August 14, 2025, from https://link.cnki.net/urlid/42.1671.TP.20250722.1357.004Hong，S. ，Zheng，X. ，Chen，J. ，Cheng，Y. ，Wang，J. ，Zhang， C. ，& Wu，C. (2023). MetaGPT：Meta programming for a multi-agent collaborative framework [EB/OL]. [2024-03-28]. http://arxiv.org/abs/2308. 00352
[22] Hong, S., Zheng, X., Chen, J., Cheng, Y., Wang, J., Zhang, C., & Wu, C. (2023). MetaGPT: Meta programming for a multi-agent collaborative framework [EB/OL]. Retrieved March 28, 2024, from http://arxiv.org/abs/2308.00352
[23] Du, X. P., & Wang, Y. Y. (2025). Research on the construction of intelligent tutoring systems empowered by retrieval-augmented generation: Based on local large models and private knowledge bases. China Educational Technology, (05), 117-127.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Journal of Education and Educational Research
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
