arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.08111cs.CLcs.AIcs.LG
收藏

语料库质量对低资源语言真的重要吗?

Does Corpus Quality Really Matter for Low-Resource Languages?

Meta AI;HiTZ Center, University of the Basque Country (UPV/EHU)(巴斯克大学HiTZ中心)

Mikel Artetxe, Itziar Aldabe, Rodrigo Agerri, Olatz Perez-de-Viñaspre, Aitor Soroa

摘要:

绝大多数非英语语料库都来自自动过滤的CommonCrawl版本。虽然先前的工作已经确定了关于这些数据集的质量的主要问题(Kreutzer等人,2021年),目前尚不清楚这对下游业绩的影响。以巴斯克语的表示学习为例,我们探索了定制的爬行(手动识别和抓取具有高质量内容的网站)作为过滤CommonCrawl的替代方案。我们的新语料库,称为EusCrawl,在大小上类似于流行的多语言语料库(如CC100和mC4)的巴斯克语部分,但根据本地注释器,它具有更高的质量。例如,66%的文档在EusCrawl中被评为高质量,而在mC4和CC100中,这一比例均低于33%。然而,我们在下游NLU任务上获得了类似的结果,无论用于预训练的语料库如何。我们的工作表明,低资源语言的NLU性能主要不受数据质量的限制,而语料库大小和域覆盖率等其他因素可以发挥更重要的作用。

英文摘要:

The vast majority of non-English corpora are derived from automatically filtered versions of CommonCrawl. While prior work has identified major issues on the quality of these datasets (Kreutzer et al., 2021), it is not clear how this impacts downstream performance. Taking representation learning in Basque as a case study, we explore tailored crawling (manually identifying and scraping websites with high-quality content) as an alternative to filtering CommonCrawl. Our new corpus, called EusCrawl, is similar in size to the Basque portion of popular multilingual corpora like CC100 and mC4, yet it has a much higher quality according to native annotators. For instance, 66% of documents are rated as high-quality for EusCrawl, in contrast with <33% for both mC4 and CC100. Nevertheless, we obtain similar results on downstream NLU tasks regardless of the corpus used for pre-training. Our work suggests that NLU performance in low-resource languages is not primarily constrained by the quality of the data, and other factors like corpus size and domain coverage can play a more important role.

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。

↑