Automated Construction of a Multi-Dialectal Saudi Corpus Using Generative Language Models
Electronics · 17.03.2026 · 15 (6) · 1241
DOI: 10.3390/electronics15061241 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronics, online erschienen am 17.03.2026. Verfasst von Khalid Almeman. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The lack of high-quality linguistic resources, especially large and diverse Arabic dialect corpora, is a major challenge in the development of Natural Language Processing (NLP) applications. By taking advantage of the generative power of Large Language Models (LLMs), this research proposes an efficient approach for the automatic construction of a large-scale corpus of Saudi dialects. We specifically translated 51,840 sentences from Modern Standard Arabic (MSA) into three major Saudi dialects: Qassim (Central), Makkah/Jeddah (Western), and Al-Ahsa (Eastern) using Google’s Gemini 1.5 Pro model. Only two items were flagged by the system as invalid outputs and removed, yielding a pipeline-level invalid output rate below 0.01%. Both quantitative and qualitative differences between MSA and its dialects were discovered through extensive linguistic analyses. Although dialectal sentences had significantly higher lexical density and type token ratios, they were always shorter and more concise. These results suggest that the generated dialectal outputs reflect expected tendencies of informal registers in this controlled, domain-specific setting, while highlighting persistent challenges for dialectal NLP—particularly orthographic variation and the lack of standardized spelling.
Zitieren
APA
Almeman, K. (2026). Automated Construction of a Multi-Dialectal Saudi Corpus Using Generative Language Models. Electronics, 15(6), 1241. https://doi.org/10.3390/electronics15061241
Vancouver
Almeman K. Automated Construction of a Multi-Dialectal Saudi Corpus Using Generative Language Models. Electronics. 2026;15(6):1241. doi:10.3390/electronics15061241
BibTeX
@article{Almeman2026,
author = {Almeman, Khalid},
title = {Automated Construction of a Multi-Dialectal Saudi Corpus Using Generative Language Models},
journal = {Electronics},
year = {2026},
volume = {15},
number = {6},
pages = {1241},
doi = {10.3390/electronics15061241}
}
EUVIMED