Agentic RAG Technology Research for Intelligent Educational Platforms
Building a next-generation intelligent educational platform that fuses Agentic RAG technology with Socratic dialogue methods
This is an Industry-Academia joint technology development project carried out together with Saltlux, Yonsei University, and KAIST as part of the RISE (Regional Innovation System & Education) program.
Research Objectives
By building a next-generation intelligent educational platform that fuses Agentic RAG technology with Socratic dialogue methods, we aim to improve hallucination issues and enhance learners’ critical thinking. We will innovate the Jeju region’s specialized industry-academia cooperative education ecosystem and provide a 24-hour AI tutoring service.
Core Technologies
- Design and development of Multi-Agent-based Agentic RAG architecture
- Construction of a Socratic dialogue management system
- Knowledge Tracing-based personalized learning support
- Constitutional AI-based educational ethics safeguards
- Development of simulation learning environments linked to Jeju’s strategic industries
- Construction of an agent reasoning evaluation framework and benchmarks
Participating Institutions
| Institution | Role |
|---|---|
| Cheju Halla University (Lead) | Research management, platform architecture design, industry-academia cooperation program operation |
| Saltlux | Agentic RAG development, FCG system, Knowledge Tracing |
| Yonsei University | Socratic dialogue LLM, reasoning pattern analysis |
| KAIST | Agent evaluation metrics, benchmark dataset construction |
Cheju Halla University Student Researchers
Six student researchers from the Department of Artificial Intelligence participate in this project, conducting research on core technologies for the AI educational platform alongside their supervising professor.
| Name | Role |
|---|---|
| 문재현 | Student Researcher |
| 김재은 | Student Researcher |
| 김웅빈 | Student Researcher |
| 홍성관 | Student Researcher |
| 김현우 | Student Researcher |
| 김호준 | Student Researcher |
Student researchers participate in reviewing the latest NLP/LLM papers through regular meetings, developing Socratic dialogue system prototypes, building simulation benchmarks for incomplete student agents, and conducting experiments on AI educational tutoring systems.
Project Information
- Research Period: August 2025 – December 2027 (29 months)
- Total R&D Budget: Approximately 1.7 billion KRW
- Principal Investigator (PI): Professor Young Joon Lee, Department of Artificial Intelligence, Cheju Halla University