OSACT 2024 Task 2: Arabic Dialect to MSA Translation

Authors

Hanin Atwany, Mohamed Bin Zayed University of Artificial IntelligenceFollow
Nour Rabih, Mohamed Bin Zayed University of Artificial IntelligenceFollow
Ibrahim Mohammed, Mohamed Bin Zayed University of Artificial IntelligenceFollow
Abdul Waheed, Mohamed Bin Zayed University of Artificial IntelligenceFollow
Bhiksha Raj, Mohamed Bin Zayed University of Artificial IntelligenceFollow

Document Type

Conference Proceeding

Publication Title

6th Workshop on Open-Source Arabic Corpora and Processing Tools, OSACT 2024 with Shared Tasks on Arabic LLMs Hallucination and Dialect to MSA Machine Translation at LREC-COLING 2024 - Workshop Proceedings

Abstract

We present the results of Shared Task "Dialect to MSA Translation", which tackles challenges posed by the diverse Arabic dialects in machine translation. Covering Gulf, Egyptian, Levantine, Iraqi and Maghrebi dialects, the task offers 1001 sentences in both MSA and dialects for fine-tuning, alongside 1888 blind test sentences. Leveraging GPT3.5, a state-of-the-art language model, our method achieved a BLEU score of 29.61. This endeavor holds significant implications for Neural Machine Translation (NMT) systems targeting low-resource languages with linguistic variation. Additionally, negative experiments involving fine-tuning AraT5 and No Language Left Behind (NLLB) using the MADAR Dataset resulted in BLEU scores of 10.41 and 11.96, respectively. Future directions include expanding the dataset to incorporate more Arabic dialects and exploring alternative NMT architectures to further enhance translation capabilities.

First Page

98

Last Page

103

Publication Date

1-1-2024

Recommended Citation

H. Atwany et al., "OSACT 2024 Task 2: Arabic Dialect to MSA Translation," 6th Workshop on Open-Source Arabic Corpora and Processing Tools, OSACT 2024 with Shared Tasks on Arabic LLMs Hallucination and Dialect to MSA Machine Translation at LREC-COLING 2024 - Workshop Proceedings, pp. 98 - 103, Jan 2024.