非自回归GEC标记的特征变换
Character Transformations for Non-Autoregressive GEC Tagging
摘要：
我们提出了一个基于字符的非自回归GEC方法，自动生成的字符转换。最近，校正编辑的每字分类已被证明是当前编码器-解码器GEC系统的有效的、可并行化的替代方案。我们发现，单词替换编辑可能是次优的，并导致爆炸的规则拼写，变音符号和错误的形态丰富的语言，并提出了一种方法，从GEC语料库生成字符转换。最后，我们训练字符转换模型，捷克，德国和俄罗斯，达到坚实的结果和显着的加速相比，自回归系统。源代码发布于https://github.com/ufal/wnut2021_character_transformations_gec。
英文摘要：
We propose a character-based nonautoregressive GEC approach, with automatically generated character transformations. Recently, per-word classification of correction edits has proven an efficient, parallelizable alternative to current encoder-decoder GEC systems. We show that word replacement edits may be suboptimal and lead to explosion of rules for spelling, diacritization and errors in morphologically rich languages, and propose a method for generating character transformations from GEC corpus. Finally, we train character transformation models for Czech, German and Russian, reaching solid results and dramatic speedup compared to autoregressive systems. The source code is released at https://github.com/ufal/wnut2021_character_transformations_gec.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
