Paraphrase Detection on Noisy Subtitles in Six Languages

Eetu Ilari Sjöblom; Mathias Johan Philip Creutz; Mikko Juhani Aulamo

Paraphrase Detection on Noisy Subtitles in Six Languages

Eetu Ilari Sjöblom, Mathias Johan Philip Creutz, Mikko Juhani Aulamo

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

Abstract

We perform automatic paraphrase detection on subtitle data from the Opusparcus corpus comprising six European languages: German, English, Finnish, French, Russian, and Swedish. We train two types of supervised sentence embedding models: a word-averaging (WA) model and a gated recurrent averaging network (GRAN) model. We find out that GRAN outperforms WA and is more robust to noisy training data. Better results are obtained with more and noisier data than less and cleaner data. Additionally, we experiment on other datasets, without reaching the same level of performance, because of domain mismatch between training and test data.

Original language	English
Title of host publication	Proceedings of the 2018 EMNLP Workshop W-NUT : The 4th Workshop on Noisy User-generated Text
Editors	Wei Xu, Alan Ritter, Tim Baldwin, Afshin Rahimi
Number of pages	10
Place of Publication	Stroudsburg
Publisher	The Association for Computational Linguistics
Publication date	1 Nov 2018
Pages	64-73
ISBN (Electronic)	978-1-948087-79-7
Publication status	Published - 1 Nov 2018
MoE publication type	A4 Article in conference proceedings
Event	Workshop on Noisy User-generated Text - Brussels, Belgium Duration: 1 Nov 2018 → 1 Nov 2018 Conference number: 4

Fields of Science

113 Computer and information sciences
6121 Languages

Access to Document

W-NUT20189Final published version, 287 KBLicence: CC BY

http://noisy-text.github.io/2018/pdf/W-NUT20189.pdfLicence: CC BY

Cite this

Sjöblom, E. I., Creutz, M. J. P., & Aulamo, M. J. (2018). Paraphrase Detection on Noisy Subtitles in Six Languages. In W. Xu, A. Ritter, T. Baldwin, & A. Rahimi (Eds.), Proceedings of the 2018 EMNLP Workshop W-NUT: The 4th Workshop on Noisy User-generated Text (pp. 64-73). The Association for Computational Linguistics. http://noisy-text.github.io/2018/pdf/W-NUT20189.pdf

@inproceedings{5813a3ab43664ca5a12dc711ff5ed1a1,

title = "Paraphrase Detection on Noisy Subtitles in Six Languages",

abstract = "We perform automatic paraphrase detection on subtitle data from the Opusparcus corpus comprising six European languages: German, English, Finnish, French, Russian, and Swedish. We train two types of supervised sentence embedding models: a word-averaging (WA) model and a gated recurrent averaging network (GRAN) model. We find out that GRAN outperforms WA and is more robust to noisy training data. Better results are obtained with more and noisier data than less and cleaner data. Additionally, we experiment on other datasets, without reaching the same level of performance, because of domain mismatch between training and test data.",

keywords = "113 Computer and information sciences, 6121 Languages",

author = "Sj{\"o}blom, {Eetu Ilari} and Creutz, {Mathias Johan Philip} and Aulamo, {Mikko Juhani}",

year = "2018",

month = nov,

day = "1",

language = "English",

pages = "64--73",

editor = "Wei Xu and Alan Ritter and Tim Baldwin and Afshin Rahimi",

booktitle = "Proceedings of the 2018 EMNLP Workshop W-NUT",

publisher = "The Association for Computational Linguistics",

address = "United States",

note = "Workshop on Noisy User-generated Text, W-NUT ; Conference date: 01-11-2018 Through 01-11-2018",

}

Sjöblom, EI, Creutz, MJP & Aulamo, MJ 2018, Paraphrase Detection on Noisy Subtitles in Six Languages. in W Xu, A Ritter, T Baldwin & A Rahimi (eds), Proceedings of the 2018 EMNLP Workshop W-NUT: The 4th Workshop on Noisy User-generated Text. The Association for Computational Linguistics, Stroudsburg, pp. 64-73, Workshop on Noisy User-generated Text, Brussels, Belgium, 01/11/2018. <http://noisy-text.github.io/2018/pdf/W-NUT20189.pdf>

Paraphrase Detection on Noisy Subtitles in Six Languages. / Sjöblom, Eetu Ilari; Creutz, Mathias Johan Philip ; Aulamo, Mikko Juhani.
Proceedings of the 2018 EMNLP Workshop W-NUT: The 4th Workshop on Noisy User-generated Text. ed. / Wei Xu; Alan Ritter; Tim Baldwin; Afshin Rahimi. Stroudsburg: The Association for Computational Linguistics, 2018. p. 64-73.

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

TY - GEN

T1 - Paraphrase Detection on Noisy Subtitles in Six Languages

AU - Sjöblom, Eetu Ilari

AU - Creutz, Mathias Johan Philip

AU - Aulamo, Mikko Juhani

N1 - Conference code: 4

PY - 2018/11/1

Y1 - 2018/11/1

N2 - We perform automatic paraphrase detection on subtitle data from the Opusparcus corpus comprising six European languages: German, English, Finnish, French, Russian, and Swedish. We train two types of supervised sentence embedding models: a word-averaging (WA) model and a gated recurrent averaging network (GRAN) model. We find out that GRAN outperforms WA and is more robust to noisy training data. Better results are obtained with more and noisier data than less and cleaner data. Additionally, we experiment on other datasets, without reaching the same level of performance, because of domain mismatch between training and test data.

AB - We perform automatic paraphrase detection on subtitle data from the Opusparcus corpus comprising six European languages: German, English, Finnish, French, Russian, and Swedish. We train two types of supervised sentence embedding models: a word-averaging (WA) model and a gated recurrent averaging network (GRAN) model. We find out that GRAN outperforms WA and is more robust to noisy training data. Better results are obtained with more and noisier data than less and cleaner data. Additionally, we experiment on other datasets, without reaching the same level of performance, because of domain mismatch between training and test data.

KW - 113 Computer and information sciences

KW - 6121 Languages

M3 - Conference contribution

SP - 64

EP - 73

BT - Proceedings of the 2018 EMNLP Workshop W-NUT

A2 - Xu, Wei

A2 - Ritter, Alan

A2 - Baldwin, Tim

A2 - Rahimi, Afshin

PB - The Association for Computational Linguistics

CY - Stroudsburg

T2 - Workshop on Noisy User-generated Text

Y2 - 1 November 2018 through 1 November 2018

ER -