[ Skip to the content ]

Institute of Formal and Applied Linguistics

at Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic

[ Back to the navigation ]


Year 2017
Type article
Status published
Language English
Author(s) Rikters, Matiss Fishel, Mark Bojar, Ondřej
Title Visualizing Neural Machine Translation Attention and Confidence
Czech title Zobrazování pozornosti a spolehlivosti neuronového strojového překladu
Journal The Prague Bulletin of Mathematical Linguistics
Publisher's city and country Prague, Czech Republic
Volume 109
Pages range 39-50
Month October
How published print
URL https://ufal.mff.cuni.cz/pbml/109/art-rikters-fishel-bojar.pdf
Supported by 2013-2017 IC1207 (PARSEME -- Parsing and multi-word expressions. Towards linguistic precision and computational efficiency in natural language processing (COST Action)) 2015-2018 H2020-ICT-2014-1-645452 (QT21: Quality Translation 21) 2017-2021 PROGRES Q18 (Společenské vědy: od víceoborovosti k mezioborovosti) 2017-2021 PROGRES Q48 (Informatika)
Czech abstract V článku popisuje nástroj pro vykreslování výstupu a pozornostních vah neuronového překladu a pro odhad spolehlivosti překladu vypočtený na základě pozornosti.
English abstract In this article, we describe a tool for visualizing the output and attention weights of neural machine translation systems and for estimating confidence about the output based on the attention. Our aim is to help researchers and developers better understand the behaviour of their NMT systems without the need for any reference translations. Our tool includes command line and web-based interfaces that allow to systematically evaluate translation outputs from various engines and experiments. We also present a web demo of our tool with examples of good and bad translations: http://ej.uz/nmt-attention
Specialization linguistics ("jazykověda")
Confidentiality default – not confidential
Open access no
DOI https://doi.org/10.1515/pralin-2017-0037
ISSN* 0032-6585
Institution* Univerzita Karlova v Praze
Creator: Common Account
Created: 11/9/17 11:33 AM
Modifier: Almighty Admin
Modified: 2/27/18 10:15 PM

Content, Design & Functionality: ÚFAL, 2006–2018. Page generated: Sat Feb 16 01:10:18 CET 2019

[ Back to the navigation ] [ Back to the content ]

100% OpenAIRE compliant