Adaptive AI Tutors: Enhancing Student Support through Dynamic Intent Parsing with Open-Source Models

Author

Bharti Garje, Chaudhari V. S

Abstract

 The rapid advancement of Artificial Intelligence (AI) and Large Language Models (LLMs) has created new opportunities for improving the quality of education through personalized and interactive learning experiences. However, traditional learning environments often face challenges in providing immediate assistance, individualized guidance, and continuous academic support to students. To address these challenges, this paper proposes AutoTA, a dynamic intent-based virtual teaching assistant that utilizes open-source Large Language Models to provide intelligent and context-aware academic assistance. The proposed system is designed to understand student queries, identify their underlying intent, and generate relevant responses based on the specific learning requirements. It categorizes student queries into different academic intents, such as conceptual explanations, homework assistance, syllabus-related questions, and lecture discussions. Based on the identified intent, the system dynamically selects an appropriate response generation strategy. Furthermore, the integration of Retrieval-Augmented Generation (RAG) enables the assistant to retrieve relevant information from uploaded academic materials, including lecture notes, textbooks, and course documents, thereby improving the relevance and reliability of generated responses. The use of open-source LLMs provides flexibility, scalability, and reduced dependency on proprietary AI services. AutoTA aims to enhance student engagement, support self-paced learning, and provide accessible academic assistance beyond traditional classroom hours. The proposed framework demonstrates the potential of combining intent classification, document retrieval, and open-source language models to develop an intelligent and adaptive educational support system.

Keywords

Artificial Intelligence, Large Language Models, Virtual Teaching Assistant, Intent Classification, Open-Source LLMs, Retrieval-Augmented Generation, Personalized Learning, Educational Technology.

Full Text:

Download Paper PDF

References

[1] R. Dahal, G. Murray, R. Chataut, M. Hefeida, A. K. Srivastava, and P. K. Gyawali, “AutoTA: A dynamic intent-based virtual teaching assistant for students using open source LLMs,” IEEE Access, vol. 13, pp. 118122–118134, 2025, doi: 10.1109/ACCESS.2025.3576329.

[2] B. Alsafari, E. Atwell, A. Walker, and M. Callaghan, “Towards effective teaching assistants: From intent-based chatbots to LLM-powered teaching assistants,” Natural Language Processing Journal, vol. 8, Art. no. 100101, 2024, doi: 10.1016/j.nlp.2024.100101.

[3] “Towards highly adaptive Edu-Chatbot,” Procedia Computer Science, vol. 198, pp. 397–403, 2022, doi: 10.1016/j.procs.2021.12.260.

[4] 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. [Online]. Available: https://arxiv.org/abs/2005.11401

[5] V. Karpukhin et al., “Dense passage retrieval for open-domain question answering,” in Proc. Conf. Empirical Methods in Natural Language Processing (EMNLP), 2020, pp. 6769–6781, doi: 10.18653/v1/2020.emnlp-main.550.

[6] J. Swacha and M. Gracel, “Retrieval-augmented generation (RAG) chatbots for education: A survey of applications,” Applied Sciences, vol. 15, no. 8, Art. no. 4234, 2025, doi: 10.3390/app15084234.

[7] Y. Yan et al., “Practical and ethical challenges of large language models in education: A systematic scoping review,” British Journal of Educational Technology, vol. 55, no. 1, pp. 90–112, 2024.

[8] S. Kweon, S. Nam, H. Lim, H. Hong, and E. Choi, “A large-scale real-world evaluation of an LLM-based virtual teaching assistant,” in Proc. 63rd Annu. Meeting Assoc. Comput. Linguistics (ACL), Industry Track, 2025, pp. 850–864, doi: 10.18653/v1/2025.acl-industry.60.

[9] D. Alexander and A. P. de Vries, “In a few words: Comparing weak supervision and LLMs for short query intent classification,” in Proc. 48th Int. ACM SIGIR Conf. Research and Development in Information Retrieval, 2025, pp. 2977–2981, doi: 10.1145/3726302.3730213

Share your valuable work from Social Media Buttons