Arabic Government Documents Corpus

Government-RestrictedGov ReuseStatisticsJSONL268K documents22 GB

86Data Health

520 downloads

Updated monthly. Last update 8 days ago. Next expected in 22 days.

About this dataset

Machine-readable corpus of Arabic government circulars, tender notices, and administrative decisions with section structure, entity mentions, and document-type labels. OCR output is human-corrected for a 40K-document gold subset. Core pre-training and retrieval source for Arabic government NLP.

Usage rights

Restricted to government entities and licensed research partners; documents referencing named individuals are filtered or redacted in line with the PDPPL before inclusion.

Lineage

Aggregated from entity document management systems and the official gazette, OCR-processed with human correction on the gold subset, and enriched with NER-based entity spans.

#arabic-nlp#documents#corpus#ner#rag#government

4 models trained on this dataset

hbku/Arabic Gov NER

Arabic NLP · 88K calls/30d

mcit/Gov Document Summarizer

Arabic NLP · 96K calls/30d

wahaj-labs/Legal Clause Retriever

RAG · 42K calls/30d

mcit/Fanar Gov 7B Instruct

RAG · 520K calls/30d

Owning entity

mcMCIT

Updated: 2026-07-22
Cadence: Monthly
Format: JSONL
Latest version: v2.4