- Published on
Tonative: Community-Driven Extension of African Datasets Through Human-AI Collaboration
- Authors
- Name
- Sharon Ibejih
- X
- Name
- Cynthia Amol
- X
Presented at Women in Machine Learning Workshop @ NeurIPS 2025
Abstract
Recent efforts to create more language resources for Africa have been led by local communities. While their efforts have resulted in improved representation for major African languages, significant gaps remain for the 2,000+ African languages that are still severely low-resource. Additionally, these initiatives have highlighted a critical issue around sustainability and continuity, as research efforts are often short-lived. In order to improve the continuity of African datasets, encouraging open contributions is essential. Open collaboration allows datasets to extend far beyond their original authors’ intended scope and enables the incorporation of language evolution over time.
Dataset curation or contributions often face significant challenges due to the high costs and scalability issues associated with human effort. Community-driven efforts produce high-quality, realistic data but suffer from scalability challenges...