Design and Implementation of an Intelligent Learning Assistance System: A Full-Stack Web Application with React and FastAPI

Qianjing Huang; Rui Jiang; Lijie Huang; Danna Li; Jieshen Huang; Jinhui Liang

doi:10.63313/JCSFT.9089

Authors

Qianjing Huang Faculty of Mathematical Sciences and Statistics, Baise University, Baise, Guengxi, 533000, China Author
Rui Jiang Faculty of Information Engineering, Baise University, Baise, Guengxi, 533000, China Author
Lijie Huang Faculty of Information Engineering, Baise University, Baise, Guengxi, 533000, China Author
Danna Li Faculty of Information Engineering, Baise University, Baise, Guengxi, 533000, China Author
Jieshen Huang Faculty of Information Engineering, Baise University, Baise, Guengxi, 533000, China Author
Jinhui Liang Faculty of Information Engineering, Baise University, Baise, Guengxi, 533000, China Author

DOI:

https://doi.org/10.63313/JCSFT.9089

Keywords:

Software Architecture, Full-Stack Development, React, Fastapi, Intelligent Learning System, Web Application, Machine Learning Integration

Abstract

This study presents the design and implementation of an Intelligent Learning Assistance System (ILAS), a full-stack web application that integrates machine learning services for student risk prediction, profiling, knowledge tracing, and personalized recommendations. The system adopts a four-tier architecture comprising a React 18 frontend with TypeScript, an Nginx gateway, a FastAPI backend with SQLAlchemy ORM, and a dedicated AI model service layer. The frontend implements 14 page routes with ECharts-based visualizations. The backend exposes 45+ RESTful API endpoints across six service modules, with JWT-based authentication and role-based access control. Key technical contributions include a Singleton Model Manager pattern for heterogeneous AI model lifecycle management, a three-strategy recommendation engine combining BKT-based weakness targeting (weight 0.5), BERT semantic matching (weight 0.3), and collaborative filtering (weight 0.2), and an asynchronous Celery-based LLM report generation pipeline. Performance evaluation demonstrates that database indexing improves query performance by 78% to 92%, and the system maintains an average response time of 312ms under 150 concurrent users with a 0.12% error rate.

References

[1] Baker R S J d, Inventado P S. Educational Data Mining and Learning Analytics[M]//Learning Analytics: From Research to Practice. New York: Springer, 2014: 61-75.

[2] Romero C, Ventura S. Educational Data Mining: A Review of the State of the Art[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part C, 2010, 40(6): 601-618.

[3] Yuan S. Research on the Design of Adaptive Learning System Based on Artificial Intelligence Driven Learning[C]//Proceedings of the 2025 2nd International Conference on Informatics Education and Computer Technology Applications (IECA 2025). Kuala Lumpur: ACM, 2025: 1-7.

[4] Chen M, Huang J, Chen W. Big Data-Driven Personalized Learning Path Recommendation via Knowledge Graph and Multi-Source Educational Data Fusion[C]//Proceedings of the 2025 3rd International Conference on Information Education and Artificial Intelligence (ICIEAI 2025). ACM, 2026: 675-679.

[5] Fielding R T. Architectural Styles and the Design of Network-based Software Architectures[D]. Irvine: University of California, 2000.

[6] Schwartz B, Zaitsev P, Tkachenko V. High Performance MySQL[M]. 4th ed. Sebastopol: O'Reilly Media, 2022.

[7] Carlson J L. Redis in Action[M]. Shelter Island: Manning Publications, 2013.

[8] Sculley D, Holt G, Golovin D, et al. Hidden Technical Debt in Machine Learning Systems[C]//Advances in Neural Information Processing Systems 28. Montreal: NIPS, 2015: 2503-2511.

[9] Ricci F, Rokach L, Shapira B. Recommender Systems Handbook[M]. 3rd ed. New York: Springer, 2022.

[10] Shi H, Zhang N, Caskurlu S, et al. Applications of Machine Learning for at-Risk Student Prediction in Online Education: A 10-Year Systematic Review of Literature[J]. Journal of Computer Assisted Learning, 2025, 41(3): e70058.