Papers by Shinya Iizuka
JMultiWOZ: A Large-Scale Japanese Multi-Domain Task-Oriented Dialogue Dataset (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing datasets for task-oriented dialogue systems in English are limited compared to Japanese. |
| Approach: | They evaluated the dialogue state tracking and response generation capabilities of Japanese language datasets using multi-domain task-oriented dialogues. |
| Outcome: | The proposed dataset provides a benchmark that is on par with MultiWOZ2.2 and the latest large language model (LLM)-based methods. |