Elamur Research⌕
☰
home / papers / 2026 / EduPulse: A Practical LLM-Enhanced Opinion Mining System for Vietnamese Student Feedback in Educational Platforms

EduPulse: A Practical LLM-Enhanced Opinion Mining System for Vietnamese Student Feedback in Educational Platforms

Paper · 2026 · Xuan Phuc Nguyen, Xuan Phi Nguyen, Vinh Tiep Nguyen et al.

Summary

Universities in Vietnam collect large amounts of student feedback. The texts are noisy, with teencode, slang, misspellings and code-switching, so manual review becomes a bottleneck. Fine-tuned models such as PhoBERT score well on benchmarks but tend to break on this kind of input unless they are retrained often. The authors built EduPulse, a deployed system with an interactive dashboard that covers four tasks: classifying overall sentiment, classifying sentiment by category, detecting suggestions, and summarizing opinions. Its design is hybrid. Prompted LLMs handle the noise-sensitive tasks (overall sentiment and suggestion identification/extraction). Aspect-level analysis uses a cheaper two-phase neural model built on cached OpenAI text-embedding-3-large vectors. Summaries come from an LLM and were judged by humans on four criteria. Running separate modules for each task scored 7.6% higher on average than one GPT-4o prompt doing everything, though the modular setup costs more compute. The paper is useful as an industry-oriented template for opinion mining in low-resource languages. Its main lessons concern robustness, maintainability and the trade-off between cost and quality.

Key points

  • Category-based analysis: a multi-label MLP first detects aspects. A second MLP then predicts sentiment from the sentence embedding concatenated with the predicted aspect-indicator vector. The authors chose this over LLM prompting and over XGBoost/SVM pipelines to get low latency and stable output.
  • Collaborative vs. monolithic: the modular architecture outperforms a single-prompt GPT-4o baseline by an average of +7.6% across sentiment, suggestion and ABSA metrics. The price is extra computation, which the authors accept for better accuracy and easier per-module maintenance.
  • Summarization evaluation: reports were rated by humans on a 1–5 scale for Informativeness, Faithfulness, Conciseness and Actionability.
    • Gemini 2.5 Flash scored highest (avg. 4.945, with 5.00 on three criteria).
    • Next came GPT-4.1 (4.830), followed by GPT-4o and Gemini 2.5 Flash Lite (both 4.820).
    • Qwen 2.5 7B scored lowest (4.255).
  • Deployment choice: Gemini 2.5 Flash is recommended for top-quality reports that do not need to be real-time. Gemini 2.5 Flash Lite is preferred for scalable, cost-sensitive operations.
  • Stated limitations: there is little benchmarking against non-LLM Vietnamese models beyond a standard PLM, and summarization depends entirely on large general-purpose LLMs. The work was funded by VNU-HCM (grant NCM2025-26-02).

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Vietnam National University Ho Chi Minh CityHo Chi Minh City · Vietnam21 works · 14 citations🎓University of Information TechnologyHo Chi Minh City · Vietnam13 works · 7 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Science Applications · 131 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,894 worksSentiment & emotion · 1,035 works

Cite

Cite as Xuan Phuc Nguyen, Xuan Phi Nguyen, Vinh Tiep Nguyen, Van Thin Dang and Ngan Luu-Thuy Nguyen. 2026. EduPulse: A Practical LLM-Enhanced Opinion Mining System for Vietnamese Student Feedback in Educational Platforms. Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 5: Industry Track).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.eacl-industry.25

@inproceedings{nguyen2026edupulse,
  title = {{EduPulse: A Practical LLM-Enhanced Opinion Mining System for Vietnamese Student Feedback in Educational Platforms}},
  author = {Xuan Phuc Nguyen, Xuan Phi Nguyen, Vinh Tiep Nguyen, Van Thin Dang and Ngan Luu-Thuy Nguyen},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 5: Industry Track)},
  year = {2026},
  doi = {10.18653/v1/2026.eacl-industry.25},
  url = {https://doi.org/10.18653/v1/2026.eacl-industry.25}
}

References

69 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Asad Abdi, Gayane Sedrakyan, Bernard Veldkamp, Jos Hillegersberg, and Stéphanie van den Berg. 2023. Students feedback analysis model using deep learning-based method and linguistic knowledge for intelligent educational systems. Soft Computing, 27:1–22. find
  2. Farhan Aftab, Sibghat Ullah Bazai, Shah Marjan, Laila Baloch, Saad Aslam, Angela Amphawan, and Tse Kian Neo. 2023. A comprehensive survey on sentiment analysis techniques. International Journal of Technology, 14(6):1288–1298. find
  3. Walid Mohamed Aly, Taysir Hassan A. Soliman, and Amr Mohamed AbdelAziz. 2025. An evaluation of large language models on text summarization tasks using prompt engineering techniques. Preprint, arXiv:2507.05123. — [arXiv:2507.05123](https://arxiv.org/abs/2507.05123) find
  4. L Godlin Atlas, Daniel Arockiam, Arvindhan Muthusamy, Balamurugan Balusamy, Shitharth Selvarajan, Taher Al-Shehari, and Nasser A Alsadhan. 2025. A modernized approach to sentiment analysis of product reviews using bigru and rnn based lstm deep learning models. Scientific Reports, 15(1):16642. find
  5. Daswin De Silva and Damminda Alahakoon. 2022. An artificial intelligence life cycle: From conception to production. Patterns, 3(6). find
  6. Abdellah Ibrahim Mohammed Elfeky, Thouqan Saleem Yakoub Masadeh, and Marwa Yasien Helmy Elbyaly. 2020. Advance organizers in flipped classroom via e-learning management system and the promotion of integrated science process skills. Thinking Skills and Creativity, 35:100622. find
  7. Brittney Exline, Melanie Duffin, Brittany Harbison, Chrissa da Gomez, and David Joyner. 2025. Using sentiment analysis to investigate peer feedback by native and non-native english speakers. Preprint, arXiv:2507.22924.
  8. Kathryn Fuller, Kathryn Morbitzer, Jacqueline Zeeman, Adam Persky, Amanda Savage, and Jacqueline McLaughlin. 2024. Exploring the use of chatgpt to analyze student course evaluation comments. BMC Medical Education, 24. find
  9. Pawanjit Singh Ghatora, Seyed Ebrahim Hosseini, Shahbaz Pervez, Muhammad Javed Iqbal, and Nabil Shaukat. 2024. Sentiment analysis of product reviews using machine learning and pre-trained llm. Big Data and Cognitive Computing, 8(12):199. find
  10. Ilya Boytsov, Vinny DeGenova, Mikhail Balyasin, Joseph Walt, Caitlin Eusden, Marie-Claire Rochat, and Margaret Pierson. 2025. End-to-end aspect-guided review summarization at scale. Preprint, arXiv:2509.26103. find
  11. Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V. Chawla, Olaf Wiest, and Xiangliang Zhang. 2024. Large language model based multi-agents: A survey of progress and challenges. Preprint, arXiv:2402.01680.
  12. Chang Cai, Shengxin Hong, Min Ma, Haiyue Feng, Sixuan Du, Minyang Chow, Winnie Teo, Siyuan Liu, and Xiuyi Fan. 2025a. Analyzing the teaching and learning environments through student feedback at scale: a multi-agent llms framework. Education and Information Technologies, 30:21815–21847. find
  13. Yue Guo, Tal August, Gondy Leroy, Trevor Cohen, and Lucy Lu Wang. 2025. Appls: Evaluating evaluation metrics for plain language summarization. Preprint, arXiv:2305.14341. find
  14. Ali Hamdi, Ahmed Abdelmoneim Mazrou, and Mohamed Shaltout. 2024. Llm-sem: A sentiment-based student engagement metric using llms for elearning platforms. In The International Conference of Advanced Computing and Informatics, pages 145– 154. Springer. find
  15. Chang Cai, Shengxin Hong, Min Ma, Haiyue Feng, Sixuan Du, Minyang Chow, Winnie Teo, Siyuan Liu, and Xiuyi Fan. 2025b. Analyzing the teaching and learning environments through student feedback at scale: a multi-agent llms framework. Education and Information Technologies, 30:21815–21847. find
  16. Clayton Cohn, Nicole Hutchins, Tuan Le, and Gautam Biswas. 2024. A chain-of-thought prompting approach with llms for evaluating students formative assessment responses in science. Proceedings of the AAAI Conference on Artificial Intelligence, 38(21):2318223190. find
  17. Thin Van Dang, Duong Ngoc Hao, and Ngan Luu-Thuy Nguyen. 2024. A study of vietnamese sentiment classification with ensemble pre-trained language models. Vietnam Journal of Computer Science, 11(01):137–165. find
  18. Ahmed M. Darwish, Essam A. Rashed, and Ghada Khoriba. 2025. Mitigating llm hallucinations using a multi-agent framework. Information, 16(7). find
  19. Nils Constantin Hellwig, Jakob Fehle, and Christian Wolff. 2025. Exploring large language models for the generation of synthetic training samples for aspect-based sentiment analysis in low resource settings. Expert Systems with Applications, 261:125514. find
  20. Chip Huyen. 2022. Designing machine learning systems. " O’Reilly Media, Inc.". find
  21. Soufian Jebbara and Philipp Cimiano. 2016. Aspect-based sentiment analysis using a two-step neural network architecture. In Semantic Web Evaluation Challenge, pages 153–167. Springer. find
  22. Ioannis Kazlaris, Efstathios Antoniou, Konstantinos Diamantaras, and Charalampos Bratsas. 2025. From illusion to insight: A taxonomic survey of hallucination mitigation techniques in llms. AI, 6(10). find
  23. Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. 2023. Large language models are zero-shot reasoners. Preprint, arXiv:2205.11916.
  24. Dat Quoc Nguyen and Anh Tuan Nguyen. 2020. PhoBERT: Pre-trained language models for Vietnamese. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 1037–1042.
  25. Anna Koufakou. 2024. Deep learning for opinion mining and topic classification of course reviews. Education and Information Technologies, 29(3):2973– 2997. find
  26. Dung Ha Nguyen, Anh Thi Hoang Nguyen, and Kiet Van Nguyen. 2024. A weakly supervised data labeling framework for machine lexical normalization in vietnamese social media. Preprint, arXiv:2409.20467. — [arXiv:2409.20467](https://arxiv.org/abs/2409.20467) find
  27. Wenna Lai, Haoran Xie, Guandong Xu, and Qing Li. 2024. Rvisa: Reasoning and verification for implicit sentiment analysis. Preprint, arXiv:2407.02340. — [arXiv:2407.02340](https://arxiv.org/abs/2407.02340) find
  28. Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out, pages 74–81, Barcelona, Spain. Association for Computational Linguistics.
  29. Bing Liu. 2022. Sentiment analysis and opinion mining. Springer Nature. find
  30. Jiacheng Liu, Alisa Liu, Ximing Lu, Sean Welleck, Peter West, Ronan Le Bras, Yejin Choi, and Hannaneh Hajishirzi. 2022. Generated knowledge prompting for commonsense reasoning. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3… find
  31. Yixin Liu, Kejian Shi, Katherine He, Longtian Ye, Alexander Fabbri, Pengfei Liu, Dragomir Radev, and Arman Cohan. 2024. On learning to summarize with large language models as references. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguist…
  32. Yanying Mao, Qun Liu, and Yu Zhang. 2024. Sentiment analysis methods, applications, and challenges: A systematic literature review. Journal of King Saud University-Computer and Information Sciences, 36(4):102048. find
  33. Jennifer Meyer, Thorben Jansen, Ronja Schiller, Lucas W. Liebenow, Marlene Steinbach, Andrea Horbach, and Johanna Fleckenstein. 2024. Using llms to bring evidence-based feedback into the classroom: Ai-generated feedback increases secondary students text revision, motivation, and positive emotions. C… find
  34. Duc Do Minh, Vinh Nguyen Van, and Thang Dam Cong. 2025. Using large language models for education managements in vietnamese with low resources. Preprint, arXiv:2501.15022. find
  35. Shubhangi Mohod. 2025. Ethical and societal implication of sentiment analysis using nlp in educational feedback system. Journal of Information Systems Engineering and Management, 10:742–749. find
  36. Kiet Van Nguyen, Duc-Vu Nguyen, Phu X. V. Nguyen, Tham T. H. Truong, and Ngan Luu-Thuy Nguyen. 2018a. Uit-vsfc: Vietnamese students feedback corpus for sentiment analysis. 2018 10th International Conference on Knowledge and Systems Engineering (KSE), pages 19–24. find
  37. Kiet Van Nguyen, Vu Duc Nguyen, Phu X. V. Nguyen, Tham T. H. Truong, and Ngan Luu-Thuy Nguyen. 2018b. Uit-vsfc: Vietnamese students feedback corpus for sentiment analysis. In 2018 10th International Conference on Knowledge and Systems Engineering (KSE), pages 19–24. find
  38. Nam Nguyen, Thang Phan, Duc-Vu Nguyen, and Kiet Nguyen. 2023. ViSoBERT: A pre-trained language model for Vietnamese social media text processing. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 5191–5207, Singapore. Association for Computational Lingu… find
  39. Quy Hoang Nguyen, Minh-Van Truong Nguyen, and Kiet Van Nguyen. 2025. New benchmark dataset and fine-grained cross-modal fusion framework for vietnamese multimodal aspect-category sentiment analysis. Multimedia Systems, 31(1):4. find
  40. Michael J. Parker, Caitlin Anderson, Claire Stone, and YeaRim Oh. 2024. A large language model approach to educational survey feedback analysis. International Journal of Artificial Intelligence in Education, 35(2):444481. find
  41. Matthew E. Peters, Sebastian Ruder, and Noah A. Smith. 2019. To tune or not to tune? adapting pre-trained representations to diverse tasks. Preprint, arXiv:1903.05987.
  42. Fina Polat, Ilaria Tiddi, and Paul Groth. 2025. Testing prompt engineering methods for knowledge extraction from text. Semantic Web, 16(2):SW–243719. find
  43. Fenghua Qi, Yuxuan Gao, Meiling Wang, Tao Jiang, and Zhenhuan Li. 2024. Data mining of online teaching evaluation based on deep learning. Mathematics, 12(17):2692. find
  44. Ramteja Sajja, Yusuf Sermet, David Cwiertny, and Ibrahim Demir. 2024. Integrating ai and learning analytics for data-driven pedagogical decisions and personalized interventions in education. Preprint, arXiv:2312.09548. — [arXiv:2312.09548](https://arxiv.org/abs/2312.09548) find
  45. Ton Nu Thi Sau, Do Phuoc Sang, and Pham Thi Thu Trang. 2021. Aspect-based sentiment analysis on students feedback in vietnamese. TNU J. Sci. Technol., 226(18):48–55. find
  46. Purwo Setiawan, Arga Seta Asmara Sakti, and Dinda Safitri Ramadhani. 2025. Listening to student voices: Aspect-based sentiment analysis of academic services using bert. Jumantara Jurnal Manajemen dan Teknologi Rekayasa. find
  47. Kathrin SeSSler, Arne Bewersdorff, Claudia Nerdel, and Enkelejda Kasneci. 2025. Towards adaptive feedback with ai: Comparing the feedback quality of llms and teachers on experimentation protocols. Preprint, arXiv:2502.12842. — [arXiv:2502.12842](https://arxiv.org/abs/2502.12842) find
  48. Thanveer Shaik, Xiaohui Tao, Christopher Dann, Haoran Xie, Yan Li, and Linda Galligan. 2023a. Sentiment analysis and opinion mining on educational data: A survey. Natural Language Processing Journal, 2:100003. find
  49. Thanveer Shaik, Xiaohui Tao, Christopher Dann, Haoran Xie, Yan Li, and Linda Galligan. 2023b. Sentiment analysis and opinion mining on educational data: A survey. Natural Language Processing Journal, 2:100003. find
  50. Neeraj Anand Sharma, ABM Shawkat Ali, and Muhammad Ashad Kabir. 2025. A review of sentiment analysis: tasks, applications, and deep learning techniques. International journal of data science and analytics, 19(3):351–388. find
  51. Irum Sindhu, Sher Muhammad Daudpota, Kamal Badar, Maheen Bakhtyar, Junaid Baber, and Mohammad Nurunnabi. 2019. Aspect-based opinion mining on students feedback for faculty teaching performance evaluation. IEEE Access, 7:108729–108741. find
  52. Van Dang Thin, Duong Ngoc Hao, and Ngan Luu-Thuy Nguyen. 2023a. A systematic literature review on vietnamese aspect-based sentiment analysis. ACM Transactions on Asian and Low-Resource Language Information Processing, 22(8):1–28. find
  53. Van Dang Thin, Duong Ngoc Hao, and Ngan Luu-Thuy Nguyen. 2023b. Vietnamese sentiment analysis: An overview and comparative study of fine-tuning pretrained language models. ACM transactions on Asian and low-resource language information processing, 22(6):1–27. find
  54. Khanh Quoc Tran, Quang Phan-Minh Huynh, Oanh Thi-Hong Le, Kiet Van Nguyen, and Ngan Luu-Thuy Nguyen. 2025. Vitasa: New benchmark and methods for vietnamese targeted aspect sentiment analysis for multiple textual domains. Computer Speech & Language, 93:101800. find
  55. Congcong Wang, Paul Nulty, and David Lillis. 2020. A comparative study on word embeddings in deep learning for text classification. In Proceedings of the 4th international conference on natural language processing and information retrieval, pages 37–46. find
  56. Peisong Wang. 2024. Student sentiment analysis and classroom feedback prediction using deep learning. Applied Mathematics and Nonlinear Sciences, 9(1). find
  57. Mayur Wankhade, Annavarapu Chandra Sekhara Rao, and Chaitanya Kulkarni. 2022. A survey on sentiment analysis methods, applications, and challenges. Artificial Intelligence Review, 55(7):5731–5780. find
  58. Borui Xu, Yao Chen, Zeyi Wen, Weiguo Liu, and Evaluating small lan-Bingsheng He. 2025a. guage models for news summarization: Implications and factors influencing performance. Preprint, arXiv:2502.00641. — [arXiv:2502.00641](https://arxiv.org/abs/2502.00641) find
  59. Palak Sood, Chengyang He, Divyanshu Gupta, Yue Ning, and Ping Wang. 2024. Understanding student sentiment on mental health support in colleges using large language models. In 2024 IEEE International Conference on Big Data (BigData), pages 1865–1872. IEEE. find
  60. Hongling Xu, Yice Zhang, Qianlong Wang, and Ruifeng Xu. 2025b. DS 2 -ABSA: Dual-stream data synthesis with label refinement for few-shot aspect-based sentiment analysis. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics, pages 15460–15478. Association for Com… find
  61. Jiamin Su, Yibo Yan, Zhuoran Gao, Han Zhang, Xiang Liu, and Xuming Hu. 2025. Cafes: A collaborative multi-agent framework for multi-granular multimodal essay scoring. arXiv preprint arXiv:2505.13965.
  62. Mike Zhang, Amalie Pernille Dilling, Léon Gondelman, Niels Erik Ruan Lyngdorf, Euan D. Lindsay, and Johannes Bjerva. 2025a. Sefl: Enhancing educational assignment feedback with llm agents. Preprint, arXiv:2502.12927. — [arXiv:2502.12927](https://arxiv.org/abs/2502.12927) find
  63. Kalpa Subbaiah and Bharath Kumar Bolla. 2024. Aspect category learning and sentimental analysis using weakly supervised learning. Procedia Computer Science, 235:1246–1257. find
  64. Ting Zhang, Ivana Clairine Irsan, Ferdian Thung, and David Lo. 2025b. Revisiting sentiment analysis for software engineering in the era of large language models. ACM Transactions on Software Engineering and Methodology, 34(3):1–30. find
  65. Dang Van Thin, Duong Ngoc Hao, and Ngan Luu-Thuy Nguyen. 2024. Prompt engineering with large language models for Vietnamese sentiment classification. In Proceedings of the 38th Pacific Asia Conference on Language, Information and Computation, pages 181–192, Tokyo, Japan. Tokyo University of Foreign… find
  66. Wenxuan Zhang, Yue Deng, Bing Liu, Sinno Pan, and Lidong Bing. 2024. Sentiment analysis in the era of large language models: A reality check. In Findings of the Association for Computational Linguistics: NAACL 2024, pages 3881–3906. Association for Computational Linguistics.
  67. Xueqiao Zhang, Chao Zhang, Jianwen Sun, Jun Xiao, Yi Yang, and Yawei Luo. 2025c. Eduplanner: Llm-based multiagent systems for customized and intelligent instructional design. IEEE Trans. Learn. Technol., 18:416427. find
  68. Xiaofeng Zheng and Jian Zhang. 2025. The usage of a transformer based and artificial intelligence driven multidimensional feedback system in english writing instruction. Scientific Reports, 15. find
  69. Zixuan Zhou, Xuefei Ning, Ke Hong, Tianyu Fu, Jiaming Xu, Shiyao Li, Yuming Lou, Luning Wang, Zhihang Yuan, Xiuhong Li, Shengen Yan, Guohao Dai, Xiao-Ping Zhang, Yuhan Dong, and Yu Wang. 2024. A survey on efficient inference for large language models. Preprint, arXiv:2404.14294.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/nguyen-2026-edupulse-practical-llm-enhanced-opinion/