Title
Back-translation as strategy to tackle the lack of corpus in natural language generation from semantic representations
Date Issued
01 January 2019
Access level
metadata only access
Resource Type
conference paper
Author(s)
University of São Paulo
Publisher(s)
Association for Computational Linguistics (ACL)
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.
Start page
94
End page
103
Language
English
OCDE Knowledge area
Informática y Ciencias de la Información
Scopus EID
2-s2.0-85097971963
Resource of which it is part
MSR@EMNLP-IJCNLP 2019 - 2nd Workshop on Multilingual Surface Realisation, Proceedings
ISBN of the container
978-195073776-5
Conference
2nd Workshop on Multilingual Surface Realisation, MSR@EMNLP-IJCNLP 2019Hong Kong3 November 2019
Sponsor(s)
The authors are grateful to CAPES and USP Research Office for supporting this work, and would like to thank NVIDIA for donating the GPU. This work is part of the OPINANDO project (more details can be found in https://sites. google.com/icmc.usp.br/opinando/), and has been partly supported by the European Commission in the framework of the H2020 Programme via contracts to UPF, with the numbers 779962-RIA, 700475-IA, 7000024-RIA, and 645012–RIA.
Sources of information:
Directorio de Producción Científica
Scopus