Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations

Marco Antonio Sobrevilla Cabezudo; Simon Mille; Thiago Alexandre Salgueiro Pardo

Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations

Marco Antonio Sobrevilla Cabezudo, Simon Mille, Thiago Alexandre Salgueiro Pardo

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

8 Scopus citations

Abstract

This paper presents an exploratory study that aims to evaluate the usefulness of back-translation in Natural Language Generation (NLG) from semantic representations for non-English languages. Specifically, Abstract Meaning Representation and Brazilian Portuguese (BP) are chosen as semantic representation and language, respectively. Two methods (focused on Statistical and Neural Machine Translation) are evaluated on two datasets (one automatically generated and another one human-generated) to compare the performance in a real context. Also, several cuts according to quality measures are performed to evaluate the importance (or not) of the data quality in NLG. Results show that there are still many improvements to be made but this is a promising approach.

Original language	English
Title of host publication	MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings
Publisher	Association for Computational Linguistics (ACL)
Pages	94-103
Number of pages	10
ISBN (Electronic)	9781950737765
State	Published - 2019
Externally published	Yes
Event	2nd Workshop on Multilingual Surface Realisation, MSR@EMNLP-IJCNLP 2019 - Hong Kong, China Duration: 3 Nov 2019 → …

Publication series

Name	MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings

Conference

Conference	2nd Workshop on Multilingual Surface Realisation, MSR@EMNLP-IJCNLP 2019
Country/Territory	China
City	Hong Kong
Period	3/11/19 → …

Cite this

Cabezudo, M. A. S., Mille, S., & Pardo, T. A. S. (2019). Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations. In MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings (pp. 94-103). (MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings). Association for Computational Linguistics (ACL).

Cabezudo, Marco Antonio Sobrevilla ; Mille, Simon ; Pardo, Thiago Alexandre Salgueiro. / Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations. MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings. Association for Computational Linguistics (ACL), 2019. pp. 94-103 (MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings).

@inproceedings{a9e34663f3f14cd9a3fe65b291758a5b,

title = "Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations",

abstract = "This paper presents an exploratory study that aims to evaluate the usefulness of back-translation in Natural Language Generation (NLG) from semantic representations for non-English languages. Specifically, Abstract Meaning Representation and Brazilian Portuguese (BP) are chosen as semantic representation and language, respectively. Two methods (focused on Statistical and Neural Machine Translation) are evaluated on two datasets (one automatically generated and another one human-generated) to compare the performance in a real context. Also, several cuts according to quality measures are performed to evaluate the importance (or not) of the data quality in NLG. Results show that there are still many improvements to be made but this is a promising approach.",

author = "Cabezudo, {Marco Antonio Sobrevilla} and Simon Mille and Pardo, {Thiago Alexandre Salgueiro}",

note = "Publisher Copyright: {\textcopyright} 2019 Association for Computational Linguistics.; 2nd Workshop on Multilingual Surface Realisation, MSR@EMNLP-IJCNLP 2019 ; Conference date: 03-11-2019",

year = "2019",

language = "English",

series = "MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings",

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

pages = "94--103",

booktitle = "MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings",

}

Cabezudo, MAS, Mille, S & Pardo, TAS 2019, Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations. in MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings. MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings, Association for Computational Linguistics (ACL), pp. 94-103, 2nd Workshop on Multilingual Surface Realisation, MSR@EMNLP-IJCNLP 2019, Hong Kong, China, 3/11/19.

Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations. / Cabezudo, Marco Antonio Sobrevilla; Mille, Simon; Pardo, Thiago Alexandre Salgueiro.
MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings. Association for Computational Linguistics (ACL), 2019. p. 94-103 (MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings).

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

TY - GEN

T1 - Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations

AU - Cabezudo, Marco Antonio Sobrevilla

AU - Mille, Simon

AU - Pardo, Thiago Alexandre Salgueiro

PY - 2019

Y1 - 2019

N2 - This paper presents an exploratory study that aims to evaluate the usefulness of back-translation in Natural Language Generation (NLG) from semantic representations for non-English languages. Specifically, Abstract Meaning Representation and Brazilian Portuguese (BP) are chosen as semantic representation and language, respectively. Two methods (focused on Statistical and Neural Machine Translation) are evaluated on two datasets (one automatically generated and another one human-generated) to compare the performance in a real context. Also, several cuts according to quality measures are performed to evaluate the importance (or not) of the data quality in NLG. Results show that there are still many improvements to be made but this is a promising approach.

AB - This paper presents an exploratory study that aims to evaluate the usefulness of back-translation in Natural Language Generation (NLG) from semantic representations for non-English languages. Specifically, Abstract Meaning Representation and Brazilian Portuguese (BP) are chosen as semantic representation and language, respectively. Two methods (focused on Statistical and Neural Machine Translation) are evaluated on two datasets (one automatically generated and another one human-generated) to compare the performance in a real context. Also, several cuts according to quality measures are performed to evaluate the importance (or not) of the data quality in NLG. Results show that there are still many improvements to be made but this is a promising approach.

UR - http://www.scopus.com/inward/record.url?scp=85097971963&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85097971963

T3 - MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings

SP - 94

EP - 103

BT - MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings

PB - Association for Computational Linguistics (ACL)

T2 - 2nd Workshop on Multilingual Surface Realisation, MSR@EMNLP-IJCNLP 2019

Y2 - 3 November 2019

ER -

Cabezudo MAS, Mille S, Pardo TAS. Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations. In MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings. Association for Computational Linguistics (ACL). 2019. p. 94-103. (MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings).

Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this