Logo Lanfrica

TerrorCoder/zimbabwe-public-service-dataset

Domaine:

natural language processing

Type de record:

dataset
Créateur:
Ter
Hôte:
A multilingual (English/Shona/Ndebele), machine-readable dataset of Zimbabwean public service procedures for retrieval, QA, and local AI benchmarking. Developed for the POTRAZ AI4I 2026 Challenge (Track 1: Data) by Team Dataprenuers. # Zimbabwe Public Service Knowledge & AI Benchmark Dataset This repository contains a highly structured, machine-readable dataset designed to bridge the public information access gap in Zimbabwe by standardizing public service procedures and translating key service guidance from English into Shona and Ndebele. The dataset includes **20,000 public service procedure records** (`SVC-00001` through `SVC-20000`) spanning all **10 provinces, 60 administrative districts, 32 urban municipalities, 60 rural district councils, and 1,958 administrative wards of Zimbabwe**, backed by **100,000 multilingual QA intent benchmark pairs** across English (`en`), Shona (`sn`), and Ndebele (`nd`). --- ## Legal Compliance & Data Protection Guarantee - **Statutory Legal Compliance**: Fully compliant with the **Freedom of Information Act [Cap 10:33]**, **Copyright and Neighboring Rights Act [Cap 26:01]**, **Cyber and Data Protection Act [Cap 12:07]**, and **Official Secrets Act [Cap 11:09]**. Read our comprehensive Legal Compliance Whitepaper. - **Zero PII Certification**: Contains 0 personal phone numbers, email addresses, or private individual records. Verified by automated PII static analysis (`scripts/audit_pii.py`). - **Terms of Use**: Licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). Read our Terms of Use & Legal Disclaimer. - **Agency Correction Protocol**: Government bodies and users can submit updates via our Public Agency Correction Policy.