Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems

Satoshi Ikeda, Kazunori Komatani, Tetsuya Ogata, Hiroshi G. Okuno

研究成果: Conference contribution

3 引用 (Scopus)

抄録

We present a method of robust domain selection against out-of-grammar (OOG) utterances in multi-domain spoken dialogue systems. These utterances cause language-understanding errors because of a limited set of grammar and vocabulary of the systems, and deteriorate the domain selection. This is critical for multi-domain spoken dialogue systems to determine a system's response. We first define a topic as a domain from which the user wants to retrieve information, and estimate it as the user's intention. This topic estimation is enabled by using a large amount of sentences collected from the Web and Latent Semantic Mapping (LSM). The results are reliable even for OOG utterances. We then integrated both the topic estimation results and the dialogue history to construct a robust domain classifier against OOG utterances. The idea of integration is based on the fact that the reliability of the dialogue history is often impeded by language-understanding errors caused by OOG utterances, from which using topic estimation obtains useful information. Experimental results using 2191 utterances showed that our integrated method reduced domain selection errors by 14.3%.

元の言語English
ホスト出版物のタイトルLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ページ294-304
ページ数11
5027 LNAI
DOI
出版物ステータスPublished - 2008
外部発表Yes
イベント21st International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2008 - Wroclaw
継続期間: 2008 6 182008 6 20

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
5027 LNAI
ISSN(印刷物)03029743
ISSN(電子版)16113349

Other

Other21st International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2008
Wroclaw
期間08/6/1808/6/20

Fingerprint

Spoken Dialogue Systems
Grammar
Language
History
Vocabulary
Semantics
Classifiers
Classifier
Dialogue
Experimental Results
Estimate

ASJC Scopus subject areas

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

これを引用

Ikeda, S., Komatani, K., Ogata, T., & Okuno, H. G. (2008). Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems. : Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (巻 5027 LNAI, pp. 294-304). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻数 5027 LNAI). https://doi.org/10.1007/978-3-540-69052-8_31

Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems. / Ikeda, Satoshi; Komatani, Kazunori; Ogata, Tetsuya; Okuno, Hiroshi G.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 巻 5027 LNAI 2008. p. 294-304 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻 5027 LNAI).

研究成果: Conference contribution

Ikeda, S, Komatani, K, Ogata, T & Okuno, HG 2008, Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems. : Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 巻. 5027 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 巻. 5027 LNAI, pp. 294-304, 21st International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2008, Wroclaw, 08/6/18. https://doi.org/10.1007/978-3-540-69052-8_31
Ikeda S, Komatani K, Ogata T, Okuno HG. Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems. : Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 巻 5027 LNAI. 2008. p. 294-304. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-540-69052-8_31
Ikeda, Satoshi ; Komatani, Kazunori ; Ogata, Tetsuya ; Okuno, Hiroshi G. / Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 巻 5027 LNAI 2008. pp. 294-304 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{b707de1586514dfcabf6836848d3044d,
title = "Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems",
abstract = "We present a method of robust domain selection against out-of-grammar (OOG) utterances in multi-domain spoken dialogue systems. These utterances cause language-understanding errors because of a limited set of grammar and vocabulary of the systems, and deteriorate the domain selection. This is critical for multi-domain spoken dialogue systems to determine a system's response. We first define a topic as a domain from which the user wants to retrieve information, and estimate it as the user's intention. This topic estimation is enabled by using a large amount of sentences collected from the Web and Latent Semantic Mapping (LSM). The results are reliable even for OOG utterances. We then integrated both the topic estimation results and the dialogue history to construct a robust domain classifier against OOG utterances. The idea of integration is based on the fact that the reliability of the dialogue history is often impeded by language-understanding errors caused by OOG utterances, from which using topic estimation obtains useful information. Experimental results using 2191 utterances showed that our integrated method reduced domain selection errors by 14.3{\%}.",
author = "Satoshi Ikeda and Kazunori Komatani and Tetsuya Ogata and Okuno, {Hiroshi G.}",
year = "2008",
doi = "10.1007/978-3-540-69052-8_31",
language = "English",
isbn = "354069045X",
volume = "5027 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "294--304",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Integrating topic estimation and dialogue history for domain selection in multi-domain spoken dialogue systems

AU - Ikeda, Satoshi

AU - Komatani, Kazunori

AU - Ogata, Tetsuya

AU - Okuno, Hiroshi G.

PY - 2008

Y1 - 2008

N2 - We present a method of robust domain selection against out-of-grammar (OOG) utterances in multi-domain spoken dialogue systems. These utterances cause language-understanding errors because of a limited set of grammar and vocabulary of the systems, and deteriorate the domain selection. This is critical for multi-domain spoken dialogue systems to determine a system's response. We first define a topic as a domain from which the user wants to retrieve information, and estimate it as the user's intention. This topic estimation is enabled by using a large amount of sentences collected from the Web and Latent Semantic Mapping (LSM). The results are reliable even for OOG utterances. We then integrated both the topic estimation results and the dialogue history to construct a robust domain classifier against OOG utterances. The idea of integration is based on the fact that the reliability of the dialogue history is often impeded by language-understanding errors caused by OOG utterances, from which using topic estimation obtains useful information. Experimental results using 2191 utterances showed that our integrated method reduced domain selection errors by 14.3%.

AB - We present a method of robust domain selection against out-of-grammar (OOG) utterances in multi-domain spoken dialogue systems. These utterances cause language-understanding errors because of a limited set of grammar and vocabulary of the systems, and deteriorate the domain selection. This is critical for multi-domain spoken dialogue systems to determine a system's response. We first define a topic as a domain from which the user wants to retrieve information, and estimate it as the user's intention. This topic estimation is enabled by using a large amount of sentences collected from the Web and Latent Semantic Mapping (LSM). The results are reliable even for OOG utterances. We then integrated both the topic estimation results and the dialogue history to construct a robust domain classifier against OOG utterances. The idea of integration is based on the fact that the reliability of the dialogue history is often impeded by language-understanding errors caused by OOG utterances, from which using topic estimation obtains useful information. Experimental results using 2191 utterances showed that our integrated method reduced domain selection errors by 14.3%.

UR - http://www.scopus.com/inward/record.url?scp=48249115587&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=48249115587&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-69052-8_31

DO - 10.1007/978-3-540-69052-8_31

M3 - Conference contribution

SN - 354069045X

SN - 9783540690450

VL - 5027 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 294

EP - 304

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -