2025/08/05 by Vukosi Marivate, Isheanesu Dzingirai, Marivate, Vukosi +21
Arts and Humanities · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Lexicography and Language Studies #Translation Studies and Practices #linguistics and terminology studies
paper · pdf · doi:10.48550/arxiv.2508.03529
openalex publication_date 2025/08/05 · openalex created_date 2025/10/04 · openalex updated_date 2026/07/28
The Mafoko project systematically aggregates, digitises, and standardises fragmented multilingual terminological resources for South Africa’s official languages. Sourced from government and academic repositories, these terminologies have historically been locked in non-machine-readable formats and inaccessible structures, limiting their use for linguistic research and NLP development. This release provides the foundational Mafoko dataset, curated under the equitable Africa-centered NOODL licensing framework. Data is released in open, machine-readable formats (CSV/JSON) with provenance metadata and ISO language identifiers. Author list Vukosi Marivate (University of Pretoria; AfriDSAI; Lelapa AI) Isheanesu Dzingirai (University of Pretoria) Fiskani Banda (University of Pretoria) Richard Lastrucci (University of Pretoria) Thapelo Sindane (University of Pretoria) Keabetswe Madumo (University of Pretoria) Kayode Olaleye (University of Pretoria) Abiodun Modupe (University of Pretoria) Unarine Netshifhefhe (University of Pretoria) Herkulaas Combrink (University of the Free State) Mohlatlego Nakeng (University of Pretoria) Matome Ledwaba (University of Pretoria) Corresponding author: [email protected]