arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.13274cs.CL
收藏

OCR改进了低资源语言的机器翻译

OCR Improves Machine Translation for Low-Resource Languages

University of Michigan(密歇根大学);Meta AI;Microsoft Turing(微软图灵研究院)

Oana Ignat, Jean Maillard, Vishrav Chaudhary, Francisco Guzmán

摘要:

我们的目标是调查目前的OCR系统的性能低资源的语言和低资源的脚本。我们介绍并公开提供一个新的基准,OCR4MT,由真实和合成数据,丰富的噪音,为60个低资源的语言在低资源脚本。我们在基准测试中评估最先进的OCR系统,并分析最常见的错误。我们表明,OCR单语数据是一个有价值的资源,可以提高性能的机器翻译模型,当用于回译。然后,我们进行消融研究,以调查OCR错误如何影响机器翻译的性能,并确定什么是最低水平的OCR质量所需的单语数据是有用的机器翻译。

英文摘要:

We aim to investigate the performance of current OCR systems on low resource languages and low resource scripts. We introduce and make publicly available a novel benchmark, OCR4MT, consisting of real and synthetic data, enriched with noise, for 60 low-resource languages in low resource scripts. We evaluate state-of-the-art OCR systems on our benchmark and analyse most common errors. We show that OCR monolingual data is a valuable resource that can increase performance of Machine Translation models, when used in backtranslation. We then perform an ablation study to investigate how OCR errors impact Machine Translation performance and determine what is the minimum level of OCR quality needed for the monolingual data to be useful for Machine Translation.

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。

↑