Preserving distributional information in dialogue act classification

Quan Hung Tran; Ingrid Zukerman; Gholamreza Haffari

doi:10.18653/v1/D17-1229

Preserving distributional information in dialogue act classification

Quan Hung Tran, Ingrid Zukerman, Gholamreza Haffari

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

Abstract

This paper introduces a novel training/decoding strategy for sequence labeling.
Instead of greedily choosing a label at each time step, and using it for
the next prediction, we retain the probability distribution over the current label,
and pass this distribution to the next prediction. This approach allows us to avoid the effect of label bias and error propagation in sequence learning/decoding. Our experiments on dialogue act classification demonstrate the effectiveness of this approach. Even though our underlying neural network model is relatively simple, it outperforms more complex neural models, achieving state-of-the-art results on
the MapTask and Switchboard corpora.

Original language	English
Title of host publication	The Conference on Empirical Methods in Natural Language Processing
Subtitle of host publication	Proceedings of the Conference - September 9-11, 2017, Copenhagen, Denmark
Editors	Rebecca Hwa, Sebastian Riedel
Place of Publication	Stroudsburg PA USA
Publisher	Association for Computational Linguistics (ACL)
Pages	2151-2156
Number of pages	6
ISBN (Print)	9781945626838
DOIs	https://doi.org/10.18653/v1/D17-1229
Publication status	Published - 2017
Event	Empirical Methods in Natural Language Processing 2017 - Copenhagen, Denmark Duration: 9 Sept 2017 → 11 Sept 2017 http://www.aclweb.org/anthology/D/D17/

Conference

Conference	Empirical Methods in Natural Language Processing 2017
Abbreviated title	EMNLP 2017
Country/Territory	Denmark
City	Copenhagen
Period	9/09/17 → 11/09/17
Internet address	http://www.aclweb.org/anthology/D/D17/

Access to Document

10.18653/v1/D17-1229

244735711-oaFinal published version, 155 KB

Cite this

Tran, Q. H., Zukerman, I., & Haffari, G. (2017). Preserving distributional information in dialogue act classification. In R. Hwa, & S. Riedel (Eds.), The Conference on Empirical Methods in Natural Language Processing: Proceedings of the Conference - September 9-11, 2017, Copenhagen, Denmark (pp. 2151-2156). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/D17-1229

Tran, Quan Hung ; Zukerman, Ingrid ; Haffari, Gholamreza. / Preserving distributional information in dialogue act classification. The Conference on Empirical Methods in Natural Language Processing: Proceedings of the Conference - September 9-11, 2017, Copenhagen, Denmark. editor / Rebecca Hwa ; Sebastian Riedel. Stroudsburg PA USA : Association for Computational Linguistics (ACL), 2017. pp. 2151-2156

@inproceedings{173eef00c32e4707be774fcea352ca4f,

title = "Preserving distributional information in dialogue act classification",

abstract = "This paper introduces a novel training/decoding strategy for sequence labeling.Instead of greedily choosing a label at each time step, and using it forthe next prediction, we retain the probability distribution over the current label,and pass this distribution to the next prediction. This approach allows us to avoid the effect of label bias and error propagation in sequence learning/decoding. Our experiments on dialogue act classification demonstrate the effectiveness of this approach. Even though our underlying neural network model is relatively simple, it outperforms more complex neural models, achieving state-of-the-art results onthe MapTask and Switchboard corpora. ",

author = "Tran, {Quan Hung} and Ingrid Zukerman and Gholamreza Haffari",

year = "2017",

doi = "10.18653/v1/D17-1229",

language = "English",

isbn = "9781945626838",

pages = "2151--2156",

editor = "Rebecca Hwa and Sebastian Riedel",

booktitle = "The Conference on Empirical Methods in Natural Language Processing",

publisher = "Association for Computational Linguistics (ACL)",

note = "Empirical Methods in Natural Language Processing 2017, EMNLP 2017 ; Conference date: 09-09-2017 Through 11-09-2017",

url = "http://www.aclweb.org/anthology/D/D17/",

}

Tran, QH, Zukerman, I & Haffari, G 2017, Preserving distributional information in dialogue act classification. in R Hwa & S Riedel (eds), The Conference on Empirical Methods in Natural Language Processing: Proceedings of the Conference - September 9-11, 2017, Copenhagen, Denmark. Association for Computational Linguistics (ACL), Stroudsburg PA USA, pp. 2151-2156, Empirical Methods in Natural Language Processing 2017, Copenhagen, Denmark, 9/09/17. https://doi.org/10.18653/v1/D17-1229

Preserving distributional information in dialogue act classification. / Tran, Quan Hung; Zukerman, Ingrid ; Haffari, Gholamreza.
The Conference on Empirical Methods in Natural Language Processing: Proceedings of the Conference - September 9-11, 2017, Copenhagen, Denmark. ed. / Rebecca Hwa; Sebastian Riedel. Stroudsburg PA USA: Association for Computational Linguistics (ACL), 2017. p. 2151-2156.

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - Preserving distributional information in dialogue act classification

AU - Tran, Quan Hung

AU - Zukerman, Ingrid

AU - Haffari, Gholamreza

PY - 2017

Y1 - 2017

N2 - This paper introduces a novel training/decoding strategy for sequence labeling.Instead of greedily choosing a label at each time step, and using it forthe next prediction, we retain the probability distribution over the current label,and pass this distribution to the next prediction. This approach allows us to avoid the effect of label bias and error propagation in sequence learning/decoding. Our experiments on dialogue act classification demonstrate the effectiveness of this approach. Even though our underlying neural network model is relatively simple, it outperforms more complex neural models, achieving state-of-the-art results onthe MapTask and Switchboard corpora.

AB - This paper introduces a novel training/decoding strategy for sequence labeling.Instead of greedily choosing a label at each time step, and using it forthe next prediction, we retain the probability distribution over the current label,and pass this distribution to the next prediction. This approach allows us to avoid the effect of label bias and error propagation in sequence learning/decoding. Our experiments on dialogue act classification demonstrate the effectiveness of this approach. Even though our underlying neural network model is relatively simple, it outperforms more complex neural models, achieving state-of-the-art results onthe MapTask and Switchboard corpora.

U2 - 10.18653/v1/D17-1229

DO - 10.18653/v1/D17-1229

M3 - Conference Paper

SN - 9781945626838

SP - 2151

EP - 2156

BT - The Conference on Empirical Methods in Natural Language Processing

A2 - Hwa, Rebecca

A2 - Riedel, Sebastian

PB - Association for Computational Linguistics (ACL)

CY - Stroudsburg PA USA

T2 - Empirical Methods in Natural Language Processing 2017

Y2 - 9 September 2017 through 11 September 2017

ER -

Tran QH, Zukerman I , Haffari G. Preserving distributional information in dialogue act classification. In Hwa R, Riedel S, editors, The Conference on Empirical Methods in Natural Language Processing: Proceedings of the Conference - September 9-11, 2017, Copenhagen, Denmark. Stroudsburg PA USA: Association for Computational Linguistics (ACL). 2017. p. 2151-2156 doi: 10.18653/v1/D17-1229