A Local-First Course-Grounded Learning System Using Parent-Child Retrieval and Quantized Large Language Models

DOI: https://doi.org/10.33650/jeecom.v8i2.17200
Authors

(1) * Nyoman Sarasuartha Mahajaya   (Faculty of Informatics and Technology, Institut Bisnis dan Teknologi Indonesia)  
        Indonesia
(2)  I Dewa Gede Aristana   (Faculty of Business and Creative Design, Institut Bisnis dan Teknologi Indonesia, Denpasar, Indonesia)  
        Indonesia
(3)  Anak Agung Gede Oka Kessawa Adnyana   (Faculty of Informatics and Technology, Institut Bisnis dan Teknologi Indonesia, Denpasar, Indonesia)  
        Indonesia
(4)  I Nyoman Darma Kotama   (Department of Information and Communication Systems, Okayama University, Okayama 700-8530, Japan)  
        Japan
(5)  I Gusti Made Ngurah Desnanjaya   (Faculty of Informatics and Technology, Institut Bisnis dan Teknologi Indonesia, Denpasar, Indonesia)  
        Indonesia
(*) Corresponding Author

Abstract


Conventional Learning Management Systems (LMSs) mainly provide fixed materials and predefined activities, while cloud-based generative artificial intelligence may require external data transmission and produce responses that are not aligned with lecturer-approved course materials. This paper presents a local-first, course-grounded learning system using structured document processing, parent-child Retrieval-Augmented Generation (RAG), semantic retrieval, and quantized Large Language Models (LLMs). The system provides cited conversational responses, source-scoped quiz generation, lecturer moderation, image annotation, and interaction-history recording. Five quantized 7B and 14B models were evaluated with 20 course-grounded questions. The 7B models reached 57.14-58.95 tokens/s using approximately 5 GB of VRAM, whereas Qwen3 14B achieved the highest Faithfulness and Answer Relevancy scores of 0.882 and 0.906. The results demonstrate a trade-off between local inference performance and course-grounded response quality.

Keywords

local-first generative AI;course-grounded learning;retrieval-augmented generation;parent-child retrieval;large language model;lecturer governance



Full Text: PDF



References


Datalab, “Marker: Convert documents to Markdown, JSON, chunks, and HTML,” 2026. [Online]. Available: https://github.com/datalab-to/marker. Accessed: Jul. 30, 2026.

T. Dettmers, M. Lewis, Y. Belkada, and L. Zettlemoyer, “LLM.int8(): 8-bit matrix multiplication for transformers at scale,” in Advances in Neural Information Processing Systems, vol. 35, 2022, pp. 30318-30332.

X. Duan, F. Nwanganga, and C. Wang, “CODE-GEN: A human-in-the-loop RAG-based agentic AI system for multiple-choice question generation,” arXiv:2604.03926, 2026, doi: 10.48550/arXiv.2604.03926.

S. Elkins, E. Kochmar, J. C. K. Cheung, and I. Serban, “How useful are educational questions generated by large language models?” arXiv:2304.06638, 2023, doi: 10.48550/arXiv.2304.06638.

S. Es, J. James, L. Espinosa Anke, and S. Schockaert, “RAGAS: Automated evaluation of retrieval augmented generation,” in Proc. 18th Conf. Eur. Chapter Assoc. Comput. Linguistics: System Demonstrations, 2024, pp. 150-158, doi: 10.18653/v1/2024.eacl-demo.16.

W. Gan, Z. Qi, J. Wu, and J. C.-W. Lin, “Large language models in education: Vision and opportunities,” arXiv:2311.13160, 2023, doi: 10.48550/arXiv.2311.13160.

Y. Gao et al., “Retrieval-augmented generation for large language models: A survey,” arXiv:2312.10997, 2023, doi: 10.48550/arXiv.2312.10997.

E. Kasneci et al., “ChatGPT for good? On opportunities and challenges of large language models for education,” Learning and Individual Differences, vol. 103, p. 102274, 2023, doi: 10.1016/j.lindif.2023.102274.

LangChain, “Text splitter integrations,” 2026. [Online]. Available: https://docs.langchain.com/oss/python/integrations/splitters. Accessed: Jul. 30, 2026.

Z. Levonian, O. Henkel, C. Li, and M.-E. Postle, “Designing safe and relevant generative chats for math learning in intelligent tutoring systems,” Journal of Educational Data Mining, vol. 17, no. 1, pp. 66-97, 2025, doi: 10.5281/zenodo.14751365.

P. Lewis et al., “Retrieval-augmented generation for knowledge-intensive NLP tasks,” in Advances in Neural Information Processing Systems, vol. 33, 2020, pp. 9459-9474.

J. Lin, J. Tang, H. Tang, S. Yang, X. Dang, C. Gan, and S. Han, “AWQ: Activation-aware weight quantization for on-device LLM compression and acceleration,” Proc. Machine Learning and Systems, vol. 6, pp. 87-100, 2024.

B. Malisetty and A. J. Perez, “Evaluating quantized Llama 2 models for IoT privacy policy language generation,” Future Internet, vol. 16, no. 7, p. 224, 2024, doi: 10.3390/fi16070224.

D. T. K. Ng, C. W. Tan, and J. K. L. Leung, “Empowering student self-regulated learning and science education through ChatGPT: A pioneering pilot study,” British Journal of Educational Technology, vol. 55, no. 4, pp. 1328-1353, 2024, doi: 10.1111/bjet.13454.

Z. Nussbaum, J. X. Morris, B. Duderstadt, and A. Mulyar, “Nomic Embed: Training a reproducible long context text embedder,” arXiv:2402.01613, 2024, doi: 10.48550/arXiv.2402.01613.

Ollama, “Ollama documentation,” 2026. [Online]. Available: https://docs.ollama.com/. Accessed: Jul. 30, 2026.

pgvector, “Open-source vector similarity search for PostgreSQL,” 2026. [Online]. Available: https://github.com/pgvector/pgvector. Accessed: Jul. 30, 2026.

PostgreSQL Global Development Group, “PostgreSQL documentation,” 2026. [Online]. Available: https://www.postgresql.org/docs/. Accessed: Jul. 30, 2026.

B. Tarun, H. Du, D. Kannan, and E. F. Gehringer, “Human-in-the-loop systems for adaptive learning using generative AI,” arXiv:2508.11062, 2025, doi: 10.48550/arXiv.2508.11062.

S. Wang et al., “Large language models for education: A survey and outlook,” arXiv:2403.18105, 2024, doi: 10.48550/arXiv.2403.18105.

A. Yang et al., “Qwen3 technical report,” arXiv:2505.09388, 2025, doi: 10.48550/arXiv.2505.09388.

T. Zhang et al., “Ask NAEP: A generative AI assistant for querying assessment information,” Journal of Measurement and Evaluation in Education and Psychology, vol. 15, special issue, pp. 378-394, 2024, doi: 10.21031/epod.1548128.

T. Zheng, W. Li, J. Bai, W. Wang, and Y. Song, “Assessing the robustness of retrieval-augmented generation systems in K-12 educational question answering with knowledge discrepancies,” arXiv:2412.08985, 2024, doi: 10.48550/arXiv.2412.08985.


Dimensions, PlumX, and Google Scholar Metrics

10.33650/jeecom.v8i2.17200


Refbacks

  • There are currently no refbacks.


Copyright (c) 2026 Nyoman Sarasuartha Mahajaya, I Dewa Gede Aristana, Anak Agung Gede Oka Kessawa Adnyana, I Nyoman Darma Kotama, I Gusti Made Ngurah Desnanjaya

 
This work is licensed under a Creative Commons Attribution License (CC BY-SA 4.0)

Journal of Electrical Engineering and Computer (JEECOM)
Published by LP3M Nurul Jadid University, Indonesia, Probolinggo, East Java, Indonesia.