Hugging Face's logo

Datasets:
dsfsi
/
za-mafoko-oertb

Modalities:
Text
Formats:
json
Languages:
Tsonga
Tswana
Zulu
Size:
1K - 10K
ArXiv:
Tags:
terminology
Libraries:
Datasets
pandas
License:
Dataset card Data Studio Files
xet
 Community
za-mafoko-oertb / README.md
vukosi's picture
vukosi
Update README.md
b977021 verified
preview code
|
Raw
History Blame Contribute Delete
7.76 kB
metadata 
license: other
license_name: noodl
license_link: https://licensingafricandatasets.com/nwulite-obodo-license
language:
  - ts
  - tn
  - zu
  - xh
  - ve
  - af
  - en
  - st
  - nr
  - ss
tags:
  - terminology
size_categories:
  - 10K<n<100K

DOI arXiv

OERTB Multilingual Terminology Lists

Dataset Description

The OERTB (Open Resource Term Bank) Multilingual Terminology Lists dataset is a comprehensive collection of specialized terminology across multiple academic domains in South African languages. This dataset contains curated multilingual dictionaries and glossaries covering diverse fields including healthcare, education, technology, sports, and governance, specifically designed to support higher education contexts.

This collection is part of the broader Mafoko: South African Terminology, Lexicon, and Glossary Project, which is dedicated to the comprehensive collection, meticulous cleaning, and transformative processing of South African language terminology lists, lexicons, and glossaries. This initiative is an integral part of the broader mission of the Data Science for Social Impact (DSFSI) lab/group, which aims to liberate and openly share as many language resources as possible.

OERTB Project Purpose

The purpose of the Open Resource Term Bank (OERTB) project is to support the collaborative development and dissemination of terminological resources, and thereby promoting the use of African languages in teaching and learning at higher education institutions. More specifically, the project is aimed at promoting the African languages as vehicles for comprehending threshold concepts in academic disciplines through practices such as translanguaging – the process and practice of shuttling between students' home languages and English to facilitate comprehension, foster social cohesion and recognise students' linguistic identities.

Dataset Structure

The dataset consists of 1 JSONL files, each containing domain-specific terminology:

Files Overview

File Domain Description
oertb_data.jsonl All_OERTB Full OERTB Data

Data Format

Each JSONL file contains entries in the following structure:

{
  "id": "unique_identifier",
  "eng": "English term",
  "afr": "Afrikaans translation",
  "xho": "isiXhosa translation",
  "ssw": "siSwati translation",
  "nbl": "isiNdebele translation",
  "tsn": "Setswana translation",
  "nso": "Sepedi translation",
  "sot": "Sesotho translation",
  "ven": "Tshivenda translation",
  "tso": "Xitsonga translation",
  "eng_pos_or_descriptor": "Part of speech or descriptor in English",
  "eng_pos_or_descriptor_info": "Additional grammatical information",
  "[lang]_pos_or_descriptor": "Part of speech for each language",
  "[lang]_pos_or_descriptor_info": "Additional grammatical information for each language"
}

Languages Covered

The dataset includes terminology in 11 official South African languages:

  • English (eng)
  • Afrikaans (afr)
  • isiXhosa (xho)
  • siSwati (ssw)
  • isiNdebele (nbl)
  • Setswana (tsn)
  • Sepedi/Northern Sotho (nso)
  • Sesotho (sot)
  • Tshivenda (ven)
  • Xitsonga (tso)

Usage

Loading the Dataset 

import json

def load_oertb_terminology(file_path):
    """Load OERTB terminology from JSONL file"""
    terms = []
    with open(file_path, 'r', encoding='utf-8') as f:
        for line in f:
            terms.append(json.loads(line))
    return terms

# Example usage
oertb_terms = load_oertb_terminology('oertb_data.jsonl')

Example Applications

  • Higher Education Support: Develop terminology resources for academic disciplines
  • Translanguaging Tools: Support shuttling between home languages and English
  • Threshold Concepts: Help students comprehend complex academic concepts in their languages
  • Language Learning: Create multilingual flashcards and vocabulary builders
  • Translation Tools: Develop domain-specific translation systems
  • Educational Resources: Build terminology databases for schools and universities
  • Research: Linguistic analysis and terminology standardization studies
  • NLP Applications: Train multilingual models for South African languages

Data Quality and Provenance

The quality and accuracy of each resource are maintained by the original authors contributing to the Open Resource Term Bank (OERTB) project, ensuring the integrity and authenticity of the linguistic data. The terminology has been curated and cleared by DSFSI (Data Science for Social Impact) for public release.

The OERTB project emphasizes collaborative development of terminological resources specifically designed to support higher education contexts and promote African languages as vehicles for academic comprehension.

For any questions or clarifications regarding the content, users are encouraged to directly contact the original authors through the OERTB project channels.

Citation

Paper Citation 

@article{marivate2025mafokostructuringbuildingopen,
      title={Mafoko: Structuring and Building Open Multilingual Terminologies for South African NLP}, 
      author={Vukosi Marivate and Isheanesu Dzingirai and Fiskani Banda and Richard Lastrucci and Thapelo Sindane and Keabetswe Madumo and Kayode Olaleye and Abiodun Modupe and Unarine Netshifhefhe and Herkulaas Combrink and Mohlatlego Nakeng and Matome Ledwaba},
      year={2025},
      eprint={2508.03529},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2508.03529}, 
}

Authors and Contributors

[Placeholder for comprehensive list of authors and contributors - To be updated]

Original Authors

  • Open Resource Term Bank (OERTB) Project Leads: Adelia Carstens, Elsabé Taljard, Mbulungeni Madiba

Dataset Curation

  • Data Science for Social Impact (DSFSI)
  • [Additional contributors to be listed]

Project Information

  • Project Website: http://www.dsfsi.co.za/za-mafoko/
  • Project Name: Mafoko: South African Terminology, Lexicon, and Glossary Project
  • Organizing Institution: Data Science for Social Impact (DSFSI)

License

[License information to be specified based on DSAC and DSFSI requirements]

Contact

For questions regarding this dataset:

Acknowledgments

We acknowledge the Open Resource Term Bank (OERTB) Project contributors for their commitment to collaborative terminological resource development and their dedication to promoting African languages in higher education contexts. We also thank the Data Science for Social Impact (DSFSI) lab for their efforts in making these resources openly available to support translanguaging practices, threshold concept comprehension, and linguistic diversity in academic spaces.

The OERTB project's focus on supporting students' linguistic identities and facilitating comprehension through translanguaging practices represents a significant contribution to multilingual education in South Africa.

This dataset is part of the broader mission to liberate and openly share South African language resources, enhancing language preservation and supporting linguistic diversity in digital spaces. The OERTB project specifically aims to promote African languages as vehicles for academic comprehension and foster inclusive higher education environments through translanguaging practices.