IST-Unbabel Participation in the WMT20 Quality Estimation Shared Task
João Martinho Moura, Miguel Vera, Daan van Stigt, Fábio Kepler, André F. T. Martins
Abstract
We 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 predictorestimator architecture, and to cope with glassbox, uncertainty-based features coming from neural machine translation systems.
Topics & Concepts
Computer scienceMachine translationTransformerEstimatorTask (project management)EstimationArtificial intelligenceArchitectureMachine learningQuality (philosophy)Natural language processingStatisticsEngineeringMathematicsSystems engineeringPhilosophyEpistemologyVisual artsElectrical engineeringArtVoltageNatural Language Processing TechniquesTopic ModelingMultimodal Machine Learning Applications