Home/Research/Personalizing University Support
Personalizing University Support
Our paper benchmarking lightweight RAG systems against commercial LLMs in higher education administration. Best Paper Award in its track at ICARC 2026.
- When
- February 2026
- My part
- Co-author
- Built with
- Retrieval-augmented generation, lightweight language models
- Title
- Personalizing University Support: Benchmarking Lightweight RAG Systems Against Commercial LLMs in Higher Education Administration
- Award
- Best Paper Award, in the track "Generative AI in Education: Emerging Paradigms, Practices, and Possibilities"
- Conference
- 6th International Conference on Advanced Research in Computing (ICARC 2026), 18 and 19 February 2026, Sabaragamuwa University of Sri Lanka
What it asks
Universities answer the same administrative questions over and over, and the answers are specific to each one. The paper compares two ways of having a machine answer them: a small language model that looks things up in the university's own material before it replies (a lightweight RAG system), and the large commercial models.
Thanks
Huge thanks to Mr. Anton Jayakody for leading this work, and to my fellow co-authors.
The announcement
Related
The code from my university research on the same subject is described under University research assistant.