Guardado en:
Detalles Bibliográficos
Autores principales: Oduwole, Mardiyyah, Olajide, Oluwatosin, Suleiman, Jamiu, Hunja, Faith, Awobade, Busayo, Adebanjo, Fatimo, Akanni, Comfort, Igwe, Chinonyelum, Ododo, Peace, Omoigui, Promise, Owodunni, Abraham, Kolawole, Steven
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2509.07471
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908601304481792
author Oduwole, Mardiyyah
Olajide, Oluwatosin
Suleiman, Jamiu
Hunja, Faith
Awobade, Busayo
Adebanjo, Fatimo
Akanni, Comfort
Igwe, Chinonyelum
Ododo, Peace
Omoigui, Promise
Owodunni, Abraham
Kolawole, Steven
author_facet Oduwole, Mardiyyah
Olajide, Oluwatosin
Suleiman, Jamiu
Hunja, Faith
Awobade, Busayo
Adebanjo, Fatimo
Akanni, Comfort
Igwe, Chinonyelum
Ododo, Peace
Omoigui, Promise
Owodunni, Abraham
Kolawole, Steven
contents The linguistic diversity across the African continent presents different challenges and opportunities for machine translation. This study explores the effects of data augmentation techniques in improving translation systems in low-resource African languages. We focus on two data augmentation techniques: sentence concatenation with back translation and switch-out, applying them across six African languages. Our experiments show significant improvements in machine translation performance, with a minimum increase of 25\% in BLEU score across all six languages. We provide a comprehensive analysis and highlight the potential of these techniques to improve machine translation systems for low-resource languages, contributing to the development of more robust translation systems for under-resourced languages.
format Preprint
id arxiv_https___arxiv_org_abs_2509_07471
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Scarcity to Efficiency: Investigating the Effects of Data Augmentation on African Machine Translation
Oduwole, Mardiyyah
Olajide, Oluwatosin
Suleiman, Jamiu
Hunja, Faith
Awobade, Busayo
Adebanjo, Fatimo
Akanni, Comfort
Igwe, Chinonyelum
Ododo, Peace
Omoigui, Promise
Owodunni, Abraham
Kolawole, Steven
Computation and Language
68T50
I.7
The linguistic diversity across the African continent presents different challenges and opportunities for machine translation. This study explores the effects of data augmentation techniques in improving translation systems in low-resource African languages. We focus on two data augmentation techniques: sentence concatenation with back translation and switch-out, applying them across six African languages. Our experiments show significant improvements in machine translation performance, with a minimum increase of 25\% in BLEU score across all six languages. We provide a comprehensive analysis and highlight the potential of these techniques to improve machine translation systems for low-resource languages, contributing to the development of more robust translation systems for under-resourced languages.
title From Scarcity to Efficiency: Investigating the Effects of Data Augmentation on African Machine Translation
topic Computation and Language
68T50
I.7
url https://arxiv.org/abs/2509.07471