Character Mapping and Ad-hoc Adaptation: Edinburgh's IWSLT 2020 Open Domain Translation System

Pinzhen Chen, Nikolay Bogoychev, Ulrich Germann

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper describes the University of Edinburgh's neural machine translation systems submitted to the IWSLT 2020 open domain Japanese <-> Chinese translation task. On top of commonplace techniques like tokenisation and corpus cleaning, we explore character mapping and unsupervised decoding-time adaptation. Our techniques focus on leveraging the provided data, and we show the positive impact of each technique through the gradual improvement of BLEU.
Original languageEnglish
Title of host publicationProceedings of the 17th International Conference on Spoken Language Translation
PublisherAssociation for Computational Linguistics
Pages122-129
Number of pages8
ISBN (Print)978-1-952148-07-1
Publication statusPublished - 9 Jul 2020
Event17th International Conference on Spoken Language Translation - Virtual conference
Duration: 9 Jul 202010 Jul 2020
http://iwslt.org/doku.php

Conference

Conference17th International Conference on Spoken Language Translation
Abbreviated titleIWSLT 2020
CityVirtual conference
Period9/07/2010/07/20
Internet address

Fingerprint Dive into the research topics of 'Character Mapping and Ad-hoc Adaptation: Edinburgh's IWSLT 2020 Open Domain Translation System'. Together they form a unique fingerprint.

Cite this