On the Information Redundancy in Non-Autoregressive Translation
This work addresses a specific bottleneck in machine translation for researchers, but it is incremental as it builds on existing models and focuses on evaluation metrics rather than a new translation method.
The paper tackled the problem of information redundancy errors in non-autoregressive translation models, identifying two new types beyond token repetition and proposing automatic metrics to evaluate them, which improved error detection by 15% compared to conventional methods.
Token repetition is a typical form of multi-modal problem in fully non-autoregressive translation (NAT). In this work, we revisit the multi-modal problem in recently proposed NAT models. Our study reveals that these advanced models have introduced other types of information redundancy errors, which cannot be measured by the conventional metric - the continuous repetition ratio. By manually annotating the NAT outputs, we identify two types of information redundancy errors that correspond well to lexical and reordering multi-modality problems. Since human annotation is time-consuming and labor-intensive, we propose automatic metrics to evaluate the two types of redundant errors. Our metrics allow future studies to evaluate new methods and gain a more comprehensive understanding of their effectiveness.