JSAI2022

Presentation information

Interactive Session

General Session » Interactive Session

[4Yin2] Interactive session 2

Fri. Jun 17, 2022 12:00 PM - 1:40 PM Room Y (Event Hall)

[4Yin2-12] Utterance generation method using database search for dialogue system in a travel agency task

〇Hideki Shiroma1, Ryuichiro Higashinaka1 (1.Univ. of Nagoya)

Keywords:dialogue system, task-oriented dialogue, utterance generation, deep learning

Recent neural-based dialogue systems can provide various information in accordance with the user’s request by utilizing the results of database search for utterance generation. In this paper, we investigate a method of utterance generation using database search for a dialogue system in a travel agency task. Specifically, we utilize the dialogue data of a travel agency task between two humans, which contains the operator’s search logs and information on tourist attractions retrieved from a database, to learn an utterance generation model based on the query information and database search results by fine-tuning a pre-trained language model. We report the results of several methods using query information and search results.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.

Password