Proceedings of the 6th Computer Science Research Days, JRI 2023, 18-20 December 2023, Ouagadougou, Burkina Faso

Research Article

Neural Machine Translation for Mooré, a Low-Resource Language

Download24 downloads
  • @INPROCEEDINGS{10.4108/eai.18-12-2023.2348140,
        author={Hamed Joseph  Ouily and Aminata  Sabane and Delwende Eliane  Birba and Rodrique  Kafando and Abdoul Kader  Kabore and Tegawende F.  Bissyand\^{e}},
        title={Neural Machine Translation for Moor\^{e}, a Low-Resource Language},
        proceedings={Proceedings of the 6th Computer Science Research Days, JRI 2023, 18-20 December 2023, Ouagadougou, Burkina Faso},
        publisher={EAI},
        proceedings_a={JRI},
        year={2024},
        month={6},
        keywords={natural language processing neural machine translation low-ressource language local language},
        doi={10.4108/eai.18-12-2023.2348140}
    }
    
  • Hamed Joseph Ouily
    Aminata Sabane
    Delwende Eliane Birba
    Rodrique Kafando
    Abdoul Kader Kabore
    Tegawende F. Bissyandé
    Year: 2024
    Neural Machine Translation for Mooré, a Low-Resource Language
    JRI
    EAI
    DOI: 10.4108/eai.18-12-2023.2348140
Hamed Joseph Ouily1,*, Aminata Sabane1, Delwende Eliane Birba2, Rodrique Kafando2, Abdoul Kader Kabore2, Tegawende F. Bissyandé2
  • 1: Université Joseph KI-ZERBO
  • 2: Centre d’Excellence CITADEL, Université Virtuelle du Burkina Faso
*Contact email: hamed.ouily@gmail.com

Abstract

Natural Language Processing (NLP) is a field of artificial intelligence with the goal of enabling machines to understand human language. Neural Machine Translation (NMT) is one of the many applications of NLP and allows for the translation of a source language into a target language. NMT has made significant progress in recent years. However, most African languages, especially those in Burkina Faso, have received very little research attention in this context. In this article, we propose automated translation models for Mooré language to French based on Transformers. We obtained an average BLUE score of 44.82 for the model trained on all the data and 65.75 for the model trained only with the Jehovah’s Witnesses Bible data for the machine translation task from Mooré to French. These encouraging results may evolve as the work is still in progress.