EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

AryWiki-Instruct: A high-fidelity instruction tuning dataset for Moroccan Arabic (Darija)

Safouane Boudakkou, Abdelaaziz El Hibaoui

Data in Brief · 2026-08 · 68 · 113140

DOI: 10.1016/j.dib.2026.113140 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Data in Brief, 2026-08. Verfasst von Safouane Boudakkou und 1 weiteren Autor:innen. Beteiligte Einrichtung: Abdelmalek Essaadi University, Faculty of Science, Tetuan, Morocco. Die Arbeit nennt 5 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This article presents AryWiki-Instruct, a high-fidelity instruction tuning dataset for Moroccan Arabic (Darija), comprising 46,590 Question and Answer (QA) pairs. The dataset was derived from a snapshot of the Moroccan Arabic Wikipedia (arywiki) and generated using the Gemini-2.5-Flash model via a Context Aware batch processing architecture. The data creation process involved parsing raw XML Wikipedia dumps, filtering for script consistency and token density, and applying a Context Injection generation strategy to prevent coreference ambiguity. To ensure high information density, the raw generated output (82,500 pairs) was subjected to a rigorous automated quality assurance pipeline. This pipeline utilized a hierarchical trigger confirmation algorithm to remove repetitive administrative census noise and employed composite embedding based clustering (multilingual-e5-large) for semantic deduplication. The final dataset is formatted as a JSONL file, providing paired instructions and responses alongside their taxonomic categories. This dataset provides a native first culturally grounded resource designed to facilitate the supervised fine tuning of Large Language Models (LLMs) in Maghrebi dialects, circumventing the syntactic limitations of translated English instruction sets.

Schlagwörter

Stichwörter der Autor:innen

DarijaInstruction tuningLarge language modelsLow resource languagesNatural language processing

Autor:innen und Institutionen

  • Abdelmalek Essaadi University, Faculty of Science, Tetuan, Morocco.

Zitieren

APA

Boudakkou, S., & El Hibaoui, A. (2026). AryWiki-Instruct: A high-fidelity instruction tuning dataset for Moroccan Arabic (Darija). Data in Brief, 68, 113140. https://doi.org/10.1016/j.dib.2026.113140

Vancouver

Boudakkou S, El Hibaoui A. AryWiki-Instruct: A high-fidelity instruction tuning dataset for Moroccan Arabic (Darija). Data in Brief. 2026;68:113140. doi:10.1016/j.dib.2026.113140

BibTeX
@article{Boudakkou2026,
  author = {Boudakkou, Safouane and El Hibaoui, Abdelaaziz},
  title = {AryWiki-Instruct: A high-fidelity instruction tuning dataset for Moroccan Arabic (Darija)},
  journal = {Data in Brief},
  year = {2026},
  volume = {68},
  pages = {113140},
  doi = {10.1016/j.dib.2026.113140}
}
Kostenlos starten →