Does corpus quality really matter for low-resource languages?

• Mikel Artetxe ; Itziar Aldabe ^[1] ; Rodrigo Agerri ^[1] ; Olatz Perez-de-Viñaspre ^[1] ; Aitor Soroa ^[1]
  1. [1] Universidad del País Vasco/Euskal Herriko Unibertsitatea

     Universidad del País Vasco/Euskal Herriko Unibertsitatea

     Leioa, España

     
• Localización: InpAkta: Inpaktu sozialerako ikerlanen akta - Journal of Research for Social Impact, ISSN 2990-2479, ISSN-e 3020-8769, Nº. 0, 2023, págs. 298-307
• Idioma: inglés
• Enlaces
  • Texto Completo Ejemplar
• Dialnet Métricas: 3 Citas
• Resumen

  • 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 down-stream 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.