Chatbot pro dietní poradenství s použitím velkých jazykových modelů
| Název práce v češtině: | Chatbot pro dietní poradenství s použitím velkých jazykových modelů |
|---|---|
| Název v anglickém jazyce: | A Diet Coaching Chatbot Using Neural Language Models |
| Klíčová slova anglicky: | NLG|AI|Natural Language Processing|chatbot|healthcare|diet|nutrition|Large Language Models|NLP|Natural Language Generation |
| Akademický rok vypsání: | 2023/2024 |
| Typ práce: | diplomová práce |
| Jazyk práce: | |
| Ústav: | Ústav formální a aplikované lingvistiky (32-UFAL) |
| Vedoucí / školitel: | Simone Balloccu, Ph.D. |
| Řešitel: | skrytý - zadáno a potvrzeno stud. odd. |
| Datum přihlášení: | 26.04.2024 |
| Datum zadání: | 26.04.2024 |
| Datum potvrzení stud. oddělením: | 26.04.2024 |
| Konzultanti: | Mgr. et Mgr. Ondřej Dušek, Ph.D. |
| Zásady pro vypracování |
| Current task-oriented chatbots mostly use template-based response generation. There is a new trend of pre-trained neural language models (Zhao et al., 2023), but these have seen limited usage so far for practical applications, especially for sensitive domains such as healthcare (Dale, 2023). Previous research by Balloccu et al. (2024) produced a large training dataset for nutritional counselling, showing that language models can produce mostly safe outputs in this domain. However, the risk of hallucination (Ji et al., 2023) still exists and the effect of AI-delivered nutritional counselling using this technology is largely unknown.
The aim of this thesis is thus twofold: (1) effective fine-tuning of neural language models for the task of nutritional counselling (Lialin et al., 2023) and paraphrasing of system responses, including intrinsic evaluation of the involved models by automatic metrics (van Miltenburg, 2023; Braggaar et al., 2024); and (2) extrinsic evaluation of these models in a practical user trial. The trial will test different versions of the diet-coaching chatbot, using both templates and LLM-driven text generation. It will evaluate dietary outcomes, as well as effects on participants’ mood and engagement. |
| Seznam odborné literatury |
| Braggaar, A., Liebrecht, C., van Miltenburg, E. and Krahmer, E., 2024. Evaluating Task-oriented Dialogue Systems: A Systematic Review of Measures, Constructs and their Operationalisations. https://doi.org/10.48550/arXiv.2312.13871.
Dale, R., 2023. Navigating the text generation revolution: Traditional data-to-text NLG companies and the rise of ChatGPT. Natural Language Engineering, 29(4), pp.1188–1197. https://doi.org/10.1017/S1351324923000347. Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y., Madotto, A. and Fung, P., 2023. Survey of Hallucination in Natural Language Generation. ACM Computing Surveys, 55(12), p.248. https://doi.org/10.1145/3571730. Lialin, V., Deshpande, V. and Rumshisky, A., 2023. Scaling Down to Scale Up: A Guide to Parameter-Efficient Fine-Tuning. https://doi.org/10.48550/arXiv.2303.15647. S. Balloccu, E. Reiter, V. Kumar, D. R. Recupero, and D. Riboni, “Ask the experts: sourcing high-quality datasets for nutritional counselling through Human-AI collaboration.” arXiv, Jan. 16, 2024. doi: 10.48550/arXiv.2401.08420. van Miltenburg, E., 2023. Evaluating NLG systems: A brief introduction. https://doi.org/10.48550/arXiv.2303.16742. Zhao, W.X., Zhou, K., Li, J., Tang, T., Wang, X., Hou, Y., Min, Y., Zhang, B., Zhang, J., Dong, Z., Du, Y., Yang, C., Chen, Y., Chen, Z., Jiang, J., Ren, R., Li, Y., Tang, X., Liu, Z., Liu, P., Nie, J.-Y. and Wen, J.-R., 2023. A Survey of Large Language Models. https://doi.org/10.48550/arXiv.2303.18223. |