In-vehicle Spoken Dialogue System Using a Question-Answering Database and a Large Language Model : Collection and Analysis of Spontaneous Utterances During Driving
質問応答データベースと大規模言語モデルを用いた車室内音声対話システム ―実走行中の自由発話収集と応答可能性の分析―
- Delivery
- Available on this site
- Format
- Price
- Non-members (tax incl.)：￥1,100 Members (tax incl.)：￥880
- Publication code
- 20264560
- Paper/Info type
- Journal of Society of Automotive Engineers of Japan
Vol.80 No.9
- Pages
- 114-119(Total 6 p)
- Date of publication
- Sep 2026
- Publisher
- JSAE
- Language
- Japanese
Detailed Information
| Category(J) | ホットトピックス Translation |
|---|---|
| Category(E) | Hot Topics |
| Author(J) | 1) 李 晃伸, 2) 柄澤 光一朗, 3) 神沼 充伸 |
| Author(E) | 1) Akinobu Lee, 2) Koichiro Karasawa, 3) Atsunobu Kaminuma |
| Affiliation(J) | 1) 名古屋工業大学, 2) トヨタシステムズ, 3) 日産自動車 |
| Abstract(J) | 本稿では、質問応答データベースと大規模言語モデルを組み合わせた車室内音声対話システムを紹介する。実車走行中にWizard of Oz方式で自由発話を収集し、両方式の応答可能性を分析した結果、文脈依存質問や仮定質問で、大規模言語モデルがデータベース型応答の範囲を補えることを示した。 Translation |
| Abstract(E) | In-vehicle voice assistants are expected to support natural interaction while driving, but they must also provide reliable responses based on vehicle-specific knowledge. In this article, we introduce an in-vehicle spoken dialogue system that combines a questionanswering database and a large language model. We collected spontaneous utterances in an actual vehicle using a Wizard-of-Oz method and analyzed the response coverage of the system. The results show that the large language model can complement the databasebased system, especially for context-dependent and hypothetical questions. |
