Grammatical Error Correction in Low-Resource Scenarios
This addresses the problem of limited grammatical error correction resources for non-English languages, providing a dataset and method for researchers and practitioners in low-resource scenarios, though it is incremental as it applies existing techniques to new data.
The paper tackles grammatical error correction for low-resource languages by introducing a new Czech dataset (AKCES-GEC) and demonstrating that a Transformer neural machine translation model, when using synthetic parallel corpus, achieves state-of-the-art results on Czech, German, and Russian datasets.
Grammatical error correction in English is a long studied problem with many existing systems and datasets. However, there has been only a limited research on error correction of other languages. In this paper, we present a new dataset AKCES-GEC on grammatical error correction for Czech. We then make experiments on Czech, German and Russian and show that when utilizing synthetic parallel corpus, Transformer neural machine translation model can reach new state-of-the-art results on these datasets. AKCES-GEC is published under CC BY-NC-SA 4.0 license at https://hdl.handle.net/11234/1-3057 and the source code of the GEC model is available at https://github.com/ufal/low-resource-gec-wnut2019.