Competency-Aware Neural Machine Translation: Can Machine Translation Know its Own Translation Quality?

TL;DR

This paper introduces a competency-aware neural machine translation model that can estimate its own translation quality without external references, matching or surpassing existing quality estimation methods.

Contribution

The paper presents a novel self-estimator integrated into NMT that assesses translation quality by reconstructing source semantics, eliminating the need for reference data.

Findings

Outperforms state-of-the-art quality estimation metrics.

Demonstrates robustness and higher correlation with human judgments.

Maintains translation performance while estimating quality.

Abstract

Neural machine translation (NMT) is often criticized for failures that happen without awareness. The lack of competency awareness makes NMT untrustworthy. This is in sharp contrast to human translators who give feedback or conduct further investigations whenever they are in doubt about predictions. To fill this gap, we propose a novel competency-aware NMT by extending conventional NMT with a self-estimator, offering abilities to translate a source sentence and estimate its competency. The self-estimator encodes the information of the decoding procedure and then examines whether it can reconstruct the original semantics of the source sentence. Experimental results on four translation tasks demonstrate that the proposed method not only carries out translation tasks intact but also delivers outstanding performance on quality estimation. Without depending on any reference or annotated data typically required by state-of-the-art metric and quality estimation methods, our model yields an even higher correlation with human quality judgments than a variety of aforementioned methods, such as BLEURT, COMET, and BERTScore. Quantitative and qualitative analyses show better robustness of competency awareness in our model.