Generování odpovědí chatbota s ohledem na personu

• Thesis title in Czech: Generování odpovědí chatbota s ohledem na personu
• Thesis title in English: Persona-aware chatbot response generation
• Key words: chatboty|dialog|generování odpovědi|zpracování přirozeného jazyka|neuronové jazykové modely
• English key words: chatbots|dialogue|response generation|natural language processing|neural language models
• Academic year of topic announcement: 2022/2023
• Thesis type: diploma thesis
• Department: Institute of Formal and Applied Linguistics (32-UFAL)
• Supervisor: Mgr. et Mgr. Ondřej Dušek, Ph.D.
• Author: hidden - assigned and confirmed by the Study Dept.
• Date of registration: 22.03.2023
• Date of assignment: 22.03.2023
• Confirmed by Study dept. on: 30.03.2023

Guidelines

Chatbots, i.e. non-task-oriented dialogue systems aimed at social conversation, have rapidly advanced in recent years with the aid of neural models (Gao et al., 2019), particularly pretrained language models (LMs) pretrained on large amounts of chat data, such as online discussion forums (Zhang et al., 2019; Roller et al., 2021). However, the biggest challenge in in developing chatbots is maintaining coherent and engaging conversations (See et al., 2019), a large part of which is ensuring that the chatbot shows a consistent persona throughout its interactions with users (Li et al., 2016). To address this, the Personachat dataset (Zhang et al., 2018) was developed as a benchmark for evaluating chatbots' persona consistency, providing conversations grounded in explicit persona descriptions.
This thesis will focus on develop chatbots for the Personachat task, starting with state-of-the-art pretrained LMs and further adapting them. The aim is on ensuring consistent personas, varied outputs, and overall coherence. The thesis will explore mainly changes to data handling and model training, such as systematic hyperparameter optimization (Denkowski & Neubig, 2017), data augmentation (Dhole et al., 2021), using natural language inference (Welleck et al., 2019) or sentence embedding similarity (Reimers & Gurevych, 2019) to detect or produce persona-consistent utterances, or employing specifically trained persona discriminators (Tang et al., 2021).
The resulting chatbot will be evaluated using standard automatic fluency and diversity metrics, as well as a small-scale human evaluation experiment.

References

M. Denkowski and G. Neubig, “Stronger Baselines for Trustable Results in Neural Machine Translation,” in Proceedings of the First Workshop on Neural Machine Translation, Vancouver, Canada, Jul. 2017, pp. 18–27.
K. D. Dhole et al., “NL-Augmenter: A Framework for Task-Sensitive Natural Language Augmentation,” arXiv:2112.02721 [cs], Dec. 2021.
J. Gao, M. Galley, and L. Li, Neural Approaches to Conversational AI: Question Answering, Task-oriented Dialogues and Social Chatbots, vol. 13. Boston / Delft: now publishers, 2019. http://arxiv.org/abs/1809.08267
J. Li, M. Galley, C. Brockett, J. Gao, and B. Dolan, “A Persona-Based Neural Conversation Model,” in Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, Berlin, Germany, 2016. https://aclanthology.org/P16-1094/
M. Mesgar, E. Simpson, and I. Gurevych, “Improving Factual Consistency Between a Response and Persona Facts,” in Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, Online, Apr. 2021, pp. 549–562. https://www.aclweb.org/anthology/2021.eacl-main.44
N. Reimers and I. Gurevych, “Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks,” in 2019 Conference on Empirical Methods in Natural Language Processing (EMNLP) and 9th International Joint Conference on Natural Language Processing (IJCNLP), Hong Kong, Nov. 2019. https://aclanthology.org/D19-1410/
S. Roller et al., “Recipes for Building an Open-Domain Chatbot,” in Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, Online, Apr. 2021, pp. 300–325. doi: 10.18653/v1/2021.eacl-main.24.
A. See, S. Roller, D. Kiela, and J. Weston, “What makes a good conversation? How controllable attributes affect human judgments,” in Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), Minneapolis, Minnesota, Jun. 2019, pp. 1702–1723. doi: 10.18653/v1/N19-1170.
F. Tang, L. Zeng, F. Wang, and J. Zhou, “Persona Authentication through Generative Dialogue.” arXiv, Oct. 25, 2021. doi: 10.48550/arXiv.2110.12949.
S. Welleck, J. Weston, A. Szlam, and K. Cho, “Dialogue Natural Language Inference,” in Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Florence, Italy, Jul. 2019, pp. 3731–3741. doi: 10.18653/v1/P19-1363.
S. Zhang, E. Dinan, J. Urbanek, A. Szlam, D. Kiela, and J. Weston, “Personalizing Dialogue Agents: I have a dog, do you have pets too?,” in Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Melbourne, Australia, Jul. 2018, pp. 2204–2213. http://aclweb.org/anthology/P18-1205
Y. Zhang et al., “DIALOGPT : Large-Scale Generative Pre-training for Conversational Response Generation,” in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, Online, Jul. 2020, pp. 270–278. https://www.aclweb.org/anthology/2020.acl-demos.30