Skip to content
conftrace_
2020 EMNLP EMNLP 2020

IST-Unbabel Participation in the WMT20 Quality Estimation Shared Task

Abstract

AbstractWe present the joint contribution of IST and Unbabel to the WMT 2020 Shared Task on Quality Estimation. Our team participated on all tracks (Direct Assessment, Post-Editing Effort, Document-Level), encompassing a total of 14 submissions. Our submitted systems were developed by extending the OpenKiwi framework to a transformer-based predictor-estimator architecture, and to cope with glass-box, uncertainty-based features coming from neural machine translation systems.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Natural Language Processing

Authors

João Moura, Miguel Vera, Daan van Stigt, Fabio Kepler, André F. T. Martins

Topics

Deep Learning > Architectures > Transformers Natural Language Processing > Applications > Machine Translation Artificial Intelligence > Core AI > Natural Language Processing Machine Learning > Learning Types > Machine Translation

Keywords

neural machine translation quality estimation uncertainty estimation transformer model predictor-estimator architecture post-editing effort uncertainty-based feature glass-box feature
Download PDF

More from EMNLP 2020

All EMNLP 2020 papers →
BERGAMOT-LATTE Submissions for the WMT20 Quality Estimation Shared Task
The NiuTrans System for the WMT20 Quality Estimation Shared Task
Two-Phase Cross-Lingual Language Model Fine-Tuning for Machine Translation Quality Estimation
TMUOU Submission for WMT20 Quality Estimation Shared Task
NICT Kyoto Submission for the WMT’20 Quality Estimation Task: Intermediate Training for Domain and Task Adaptation
TransQuest at WMT2020: Sentence-Level Direct Assessment