{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,27]],"date-time":"2026-01-27T23:42:40Z","timestamp":1769557360943,"version":"3.49.0"},"reference-count":41,"publisher":"Association for Computing Machinery (ACM)","issue":"6","funder":[{"DOI":"10.13039\/100019600","name":"Hanoi University of Science and Technology","doi-asserted-by":"crossref","award":["T2024-PC-041"],"award-info":[{"award-number":["T2024-PC-041"]}],"id":[{"id":"10.13039\/100019600","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Asian Low-Resour. Lang. Inf. Process."],"published-print":{"date-parts":[[2025,6,30]]},"abstract":"<jats:p>The development of large language models (LLMs) such as ChatGPT and Gemini has led to impressive advancements in question answering (QA) systems. However, they often rely on generic knowledge from the internet, resulting in hallucinated answers when applied to domain-specific QA tasks. Furthermore, their operational dependence on powerful GPUs poses challenges for practical software deployment. Building a QA systems for low-resource languages like Vietnamese is even more challenging due to the scarcity of labeled data and limited pre-trained language models. In this study, we aim at constructing a Vietnamese legal QA system using a retrieval-augmented generation approach to reduce incorrect outputs. Our focus is on improving answer generation accuracy by training small-scale LLMs suitable for real-world deployment. Our contributions are: (i) constructing Vietnamese legal provisions and QA datasets for training the system; and (ii) proposing methods to fine-tune language models with QA capabilities in the legal domain. Experimental results demonstrate that it is possible to train an LLM with fewer computational resources and a smaller dataset while maintaining effectiveness. Our findings highlight that designing an efficient training and fine-tuning strategy is crucial for overcoming these challenges, particularly in the context of Vietnamese legal question-answering tasks.<\/jats:p>","DOI":"10.1145\/3732938","type":"journal-article","created":{"date-parts":[[2025,4,29]],"date-time":"2025-04-29T11:16:59Z","timestamp":1745925419000},"page":"1-17","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Optimizing Answer Generator in Vietnamese Legal Question Answering Systems Using Language Models"],"prefix":"10.1145","volume":"24","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7298-1598","authenticated-orcid":false,"given":"Huong","family":"Le","sequence":"first","affiliation":[{"name":"Hanoi University of Science and Technology","place":["Hanoi, Viet Nam"]}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-5521-7490","authenticated-orcid":false,"given":"Ngoc","family":"Luu","sequence":"additional","affiliation":[{"name":"Hanoi University of Science and Technology","place":["Hanoi, Viet Nam"]}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-1849-0623","authenticated-orcid":false,"given":"Thanh","family":"Nguyen","sequence":"additional","affiliation":[{"name":"Hanoi University of Science and Technology","place":["Hanoi, Viet Nam"]}]},{"ORCID":"https:\/\/orcid.org\/0009-0001-8487-2943","authenticated-orcid":false,"given":"Tuan","family":"Dao","sequence":"additional","affiliation":[{"name":"Hanoi University of Science and Technology","place":["Hanoi, Viet Nam"]}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9254-1327","authenticated-orcid":false,"given":"Sang","family":"Dinh","sequence":"additional","affiliation":[{"name":"Hanoi University of Science and Technology","place":["Hanoi, Viet Nam"]}]}],"member":"320","published-online":{"date-parts":[[2025,6,18]]},"reference":[{"key":"e_1_3_3_2_2","unstructured":"T. B. Brown B. Mann N. Ryder M. Subbiah J. Kaplan P. Dhariwal and D. Amodei. 2020. Language models are few-shot learners. In Advances in Neural Information Processing Systems 33 (2020) 1877--1901. Retrieved from https:\/\/arxiv.org\/abs\/2005.14165"},{"key":"e_1_3_3_3_2","unstructured":"Dichvucong website. 2024. Retrieved from https:\/\/dichvucong.gov.vn\/ (last visited December 2024)."},{"key":"e_1_3_3_4_2","first-page":"186","article-title":"A vietnamese question answering system in vietnam's legal documents","volume":"14","author":"Huu-Thanh Duong","year":"2014","unstructured":"Duong Huu-Thanh and Bao-Quoc Ho. 2014. A vietnamese question answering system in vietnam's legal documents. In Computer Information Systems and Industrial Management: 13th IFIP TC8 International Conference, CISIM 2014, Ho Chi Minh City, Vietnam, November 5-7, 2014. Proceedings 14, 186\u2013197. Springer Berlin.","journal-title":"Computer Information Systems and Industrial Management: 13th IFIP TC8 International Conference, CISIM 2014, Ho Chi Minh City, Vietnam"},{"key":"e_1_3_3_5_2","doi-asserted-by":"crossref","unstructured":"X. Q. Dao and N. B. Le. 2023. LLMs performance on Vietnamese high school biology examination. In International Journal of Modern Education and Computer Science 15 6 (2023) 14--30.","DOI":"10.5815\/ijmecs.2023.06.02"},{"key":"e_1_3_3_6_2","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","author":"Devlin Jacob","year":"2019","unstructured":"Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In North American Association for Computational Linguistics (NAACL).","journal-title":"North American Association for Computational Linguistics (NAACL)"},{"key":"e_1_3_3_7_2","doi-asserted-by":"publisher","unstructured":"G. Team R. Anil S. Borgeaud J. B. Alayrac J. Yu R. Soricut J. Schalkwyk A. M. Dai A. Hauth K. Millican and D. Silver. 2023. Gemini: A family of highly capable multimodal models. arXiv:2312.11805. Retrieved from 10.48550\/arXiv.2312.11805","DOI":"10.48550\/arXiv.2312.11805"},{"key":"e_1_3_3_8_2","doi-asserted-by":"publisher","unstructured":"R. Harang and H. Sanders. 2023. Catastrophic forgetting in the context of model updates. arXiv:2306.10181. Retrieved from 10.48550\/arXiv.2306.10181","DOI":"10.48550\/arXiv.2306.10181"},{"key":"e_1_3_3_9_2","volume-title":"Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence","unstructured":"W. Huang, J. Jiang, Q. Qu, and M. Yang. 2020. AILA: A question answering system in the legal domain. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. 5258--5260."},{"key":"e_1_3_3_10_2","doi-asserted-by":"publisher","unstructured":"W. Huang X. Zheng X. Ma H. Qin C. Lv H. Chen J. Luo X. Qi X. Liu M. Magno. 2024. An empirical study of LLaMA3 quantization: From LLMs to MLLMs. arXiv:2404.14047. Retrieved from 10.48550\/arXiv.2404.14047","DOI":"10.48550\/arXiv.2404.14047"},{"key":"e_1_3_3_11_2","doi-asserted-by":"publisher","unstructured":"D. Kalajdzievski. 2024. Scaling laws for forgetting when fine-tuning large language models. arXiv:2401.05605. Retrieved from 10.48550\/arXiv.2401.05605","DOI":"10.48550\/arXiv.2401.05605"},{"key":"e_1_3_3_12_2","doi-asserted-by":"crossref","unstructured":"E. Kasneci et al. 2023. ChatGPT for good? On opportunities and challenges of large language models for education. In Learning and Individual Differences 103 1 (2023) 102274.","DOI":"10.1016\/j.lindif.2023.102274"},{"key":"e_1_3_3_13_2","doi-asserted-by":"crossref","unstructured":"S. Khazaeli J. Punuru C. Morris S. Sharma B. Staub M. Cole S. Chiu-Webster D. Sakalley. 2021. A free format legal question answering system. In Proceedings of the Natural Legal Language Processing Workshop. 107--113.","DOI":"10.18653\/v1\/2021.nllp-1.11"},{"key":"e_1_3_3_14_2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.coling-main.86"},{"key":"e_1_3_3_15_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-10061-6_14"},{"key":"e_1_3_3_16_2","doi-asserted-by":"publisher","unstructured":"G. Lai Q. Xie H. Liu Y. Yang and E. Hovy. 2017. Large-scale ReAding comprehension dataset from examinations. arXiv:1704.04683. Retrieved from 10.48550\/arXiv.1704.04683","DOI":"10.48550\/arXiv.1704.04683"},{"key":"e_1_3_3_17_2","unstructured":"Lang Chain. 2024. Q&A with RAG. Retrieved from https:\/\/python.langchain.com\/v0.1\/docs\/use_cases\/question_answering\/ (last visited December 2024)."},{"key":"e_1_3_3_18_2","doi-asserted-by":"publisher","unstructured":"M. Lewis Y. Liu N. Goyal M. Ghazvininejad A. Mohamed O. Levy V. Stoyanov L. Zettlemoyer. 2019. Bart: Denoising sequence-to-sequence pre-training for natural language generation translation and comprehension. arXiv:1910.13461. Retrieved from 10.48550\/arXiv.1910.13461","DOI":"10.48550\/arXiv.1910.13461"},{"key":"e_1_3_3_19_2","doi-asserted-by":"publisher","unstructured":"A. Grattafiori et al. 2024. The llama 3 herd of models. arXiv:2407.21783. Retrieved from 10.48550\/arXiv.2407.21783","DOI":"10.48550\/arXiv.2407.21783"},{"key":"e_1_3_3_20_2","article-title":"ROUGE: A package for automatic evaluation of summaries","author":"Lin Chin-Yew","year":"2004","unstructured":"Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out, pages 74\u201381, Barcelona, Spain. Association for Computational Linguistics.","journal-title":"Text Summarization Branches Out, pages 74\u201381, Barcelona, Spain"},{"key":"e_1_3_3_21_2","doi-asserted-by":"publisher","unstructured":"H. Li F. Koto M. Wu A. F. Aji and T. Baldwin. 2023. Bactrian-X: A multilingual replicable instruction-following model with low-rank adaptation. arXiv:2305.15011. Retrieved from 10.48550\/arXiv.2305.15011","DOI":"10.48550\/arXiv.2305.15011"},{"key":"e_1_3_3_22_2","doi-asserted-by":"publisher","unstructured":"Y. Liu M. Ott N. Goyal J. Du M. Joshi D. Chen O. Levy M. Lewis L. Zettlemoyer and V. Stoyanov. 2019. RoBERTa: A robustly optimized BERT pretraining approach. arXiv:1907.11692. Retrieved from 10.48550\/arXiv.1907.11692","DOI":"10.48550\/arXiv.1907.11692"},{"key":"e_1_3_3_23_2","doi-asserted-by":"publisher","DOI":"10.1109\/NICS48868.2019.9023886"},{"key":"e_1_3_3_24_2","doi-asserted-by":"publisher","unstructured":"Q. Nguyen H. Pham and D. Dao. 2023. VinaLLaMA: LLaMA-based vietnamese foundation model. arXiv:2312.11011. Retrieved from 10.48550\/arXiv.2312.11011","DOI":"10.48550\/arXiv.2312.11011"},{"key":"e_1_3_3_25_2","doi-asserted-by":"publisher","unstructured":"T. H. Nguyen A. C. Le and V. C. Nguyen. 2024. ViLLM-Eval: A comprehensive evaluation suite for vietnamese large language models. arXiv:2404.11086. Retrieved from 10.48550\/arXiv.2404.11086","DOI":"10.48550\/arXiv.2404.11086"},{"key":"e_1_3_3_26_2","doi-asserted-by":"publisher","unstructured":"N. Q. Duc L. H. Son N. D. Nhan N. D. N. Minh L. T. Huong and D. V. Sang. 2024. Towards comprehensive vietnamese retrieval-augmented generation and large language models. arXiv:2403.01616. Retrieved from 10.48550\/arXiv.2403.01616","DOI":"10.48550\/arXiv.2403.01616"},{"key":"e_1_3_3_27_2","doi-asserted-by":"crossref","unstructured":"Y. Liu J. Gu N. Goyal X. Li S. Edunov M. Ghazvininejad M. Lewis and L. Zettlemoyer. 2020. Multilingual denoising pre-training for neural machine translation. Transactions of the Association for Computational Linguistics 8 1 (2020) 726--742.","DOI":"10.1162\/tacl_a_00343"},{"key":"e_1_3_3_28_2","doi-asserted-by":"publisher","DOI":"10.1109\/KSE59128.2023.10299515"},{"key":"e_1_3_3_29_2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-industry.70"},{"key":"e_1_3_3_30_2","doi-asserted-by":"crossref","unstructured":"A. Pham Duy and H. Le Thanh. 2023. A question-answering system for vietnamese public administrative services. In Proceedings of the 12th International Symposium on Information and Communication Technology. 85--92.","DOI":"10.1145\/3628797.3628965"},{"key":"e_1_3_3_31_2","volume-title":"Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing (EMNLP 2016)","unstructured":"P. Rajpurkar, J. Zhang, K. Lopyrev, and P. Liang. 2016. SQuAD: 100,000+ Questions for machine comprehension of text. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing (EMNLP 2016). 1--10."},{"key":"e_1_3_3_32_2","doi-asserted-by":"publisher","unstructured":"V. Sanh A.Webson C. Raffel S. H. Bach L. Sutawika Z. Alyafeai A. Chaffin A. Stiegler T. L. Scao A. Raja and M. Dey. 2021. Multitask prompted training enables zero-shot task generalization. arXiv:2110.08207. Retrieved from 10.48550\/arXiv.2110.08207","DOI":"10.48550\/arXiv.2110.08207"},{"key":"e_1_3_3_33_2","first-page":"65","article-title":"Biomedical question types classification using syntactic and rule based approach. 2015","volume":"1","author":"Sarrouti M.","year":"2015","unstructured":"M. Sarrouti, A. Lachkar, and S. E. A. Ouatik. 2015. Biomedical question types classification using syntactic and rule based approach. 2015. In 2015 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K), 1 (2015), 65\u2013272. IEEE.","journal-title":"2015 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K)"},{"key":"e_1_3_3_34_2","volume-title":"Stanford Center for Research on Foundation Models","author":"Taori Rohan","year":"2023","unstructured":"Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B. Hashimoto. 2023. Alpaca: A strong, replicable instruction-following model. Stanford Center for Research on Foundation Models 3 6 (2023), 7. Retrieved from https:\/\/crfm.stanford.edu\/2023\/03\/13\/alpaca.html"},{"key":"e_1_3_3_35_2","unstructured":"Thuvienphapluat website. Retrieved from https:\/\/thuvienphapluat.vn\/ (last visited December 2024)."},{"key":"e_1_3_3_36_2","doi-asserted-by":"publisher","DOI":"10.21437\/Interspeech.2022-10177"},{"key":"e_1_3_3_37_2","doi-asserted-by":"publisher","unstructured":"H. Touvron L. Martin K. Stone P. Albert A. Almahairi Y. Babaei N. Bashlykov S. Batra P. Bhargava S. Bhosale and D. Bikel. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv:2307.09288. Retrieved from 10.48550\/arXiv.2307.09288","DOI":"10.48550\/arXiv.2307.09288"},{"key":"e_1_3_3_38_2","unstructured":"Thi-Hai-Yen Vuong Ha-Thanh Nguyen Quang-Huy Nguyen Le-Minh Nguyen and Xuan-Hieu Phan. 2023. Improving vietnamese legal question-answering system based on automatic data enrichment. arXiv:2306.04841. Retrieved from https:\/\/arxiv.org\/abs\/2306.04841"},{"key":"e_1_3_3_39_2","doi-asserted-by":"publisher","DOI":"10.1007\/s10506-022-09319-6"},{"key":"e_1_3_3_40_2","first-page":"3","volume-title":"CCF International Conference on Natural Language Processing and Chinese Computing","author":"Xie R.","year":"2020","unstructured":"R. Xie, Y. Lu, F. Lin, and L. Lin. 2020. Faq-based question answering via knowledge anchors. 2020. In CCF International Conference on Natural Language Processing and Chinese Computing, 3\u201315. Springer."},{"key":"e_1_3_3_41_2","article-title":"Anthropic's claude instant: A smaller, faster and cheaper language model","author":"Yao Deborah.","year":"2023","unstructured":"Yao Deborah. 2023. Anthropic's claude instant: A smaller, faster and cheaper language model. AI Business. Retrieved from https:\/\/aibusiness.com\/nlp\/anthropic-s-claude-instant-a-smaller-faster-and-cheaper-language-model (last visited 18 Jun, 2024).","journal-title":"AI Business"},{"key":"e_1_3_3_42_2","doi-asserted-by":"publisher","unstructured":"S. Yue W. Chen S. Wang B. Li C. Shen S. Liu Y. Zhou Y. Xiao S. Yun W. Lin X. Huang and Z. Wei. 2023. Disc-LawLLM: Fine-tuning large language models for intelligent legal services. arXiv:2309.11325. Retrieved from 10.48550\/arXiv.2309.11325","DOI":"10.48550\/arXiv.2309.11325"}],"container-title":["ACM Transactions on Asian and Low-Resource Language Information Processing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3732938","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T01:20:27Z","timestamp":1750296027000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3732938"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6,18]]},"references-count":41,"journal-issue":{"issue":"6","published-print":{"date-parts":[[2025,6,30]]}},"alternative-id":["10.1145\/3732938"],"URL":"https:\/\/doi.org\/10.1145\/3732938","relation":{},"ISSN":["2375-4699","2375-4702"],"issn-type":[{"value":"2375-4699","type":"print"},{"value":"2375-4702","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,6,18]]},"assertion":[{"value":"2024-06-19","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-02-12","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-06-18","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}