Building Machine Translation Systems for the Next Thousand Languages
Abstract
In this paper we share findings from our effort to build practical machine translation (MT) systems capable of translating across over one thousand languages. We describe results in three research domains: (i) Building clean, web-mined datasets for 1500+ languages by leveraging semi-supervised pre-training for language identification and developing data-driven filtering techniques; (ii) Developing practical MT models for under-served languages by leveraging massively multilingual models trained with supervised parallel data for over 100 high-resource languages and monolingual datasets for an additional 1000+ languages; and (iii) Studying the limitations of evaluation metrics for these languages and conducting qualitative analysis of the outputs from our MT models, highlighting several frequent error modes of these types of models. We hope that our work provides useful insights to practitioners working towards building MT systems for currently understudied languages, and highlights research directions that can complement the weaknesses of massively multilingual models in data-sparse settings.
Cite
@article{arxiv.2205.03983,
title = {Building Machine Translation Systems for the Next Thousand Languages},
author = {Ankur Bapna and Isaac Caswell and Julia Kreutzer and Orhan Firat and Daan van Esch and Aditya Siddhant and Mengmeng Niu and Pallavi Baljekar and Xavier Garcia and Wolfgang Macherey and Theresa Breiner and Vera Axelrod and Jason Riesa and Yuan Cao and Mia Xu Chen and Klaus Macherey and Maxim Krikun and Pidong Wang and Alexander Gutkin and Apurva Shah and Yanping Huang and Zhifeng Chen and Yonghui Wu and Macduff Hughes},
journal= {arXiv preprint arXiv:2205.03983},
year = {2022}
}
Comments
V2: updated with some details from 24-language Google Translate launch in May 2022 V3: spelling corrections, additional acknowledgements