Lacuna Funded Project: MasakhaNER
April 29, 2022 · View on GitHub
Datasets developed by the projects are:
- MAFAND-MT: Masakhane Anglo & Franco African News Corpus for Machine Translation
- MasakhaNER 2.0: Named Entity Recognition datasets for 20 African languages
- African POS datasets
Team & Partners
- Peter Nabende (Makerere University) - Principal Investigator
- David Ifeoluwa Adelani (Masakhane; Saarland University) - NER Coordinator
- Bamba Dione (Masakhane; University of Bergen) - POS Coordinator
- Jade Abbott (Masakhane; Retro Rabbit) - Data & Translation Coordinator
- Constantine Lignos (Masakhane; Brandeis University) - Quality Control
- Daniel D’souza (Masakhane) - Tool Management
- Sascha Heyer (IO Annotator) - Tool Development & Support
Language Coordinators
| Language | Coordinator |
|---|---|
| Bambara | Allahsera Auguste Tapo |
| Chichewa | Amelia Taylor |
| Ewe | Godson Kalipe |
| Fon | Bonaventure Dossou |
| Ghomala | Koagne Victoire Memdjokam |
| Hausa | Tajuddeen Gwadabe |
| Igbo | Chris Emezue |
| Kinyarwanda | Happy Buzaaba |
| Luganda | Jonathan Mukiibi |
| Luo | Perez Ogayo |
| Moore | Fatoumata Kabore |
| Nigerian-Pidgin | Aremu Anuoluwapo |
| Setswana | Valencia Wagner |
| Shona | Blessing Sibanda |
| Swahili | Catherine Gitau |
| Twi | Edwin Buabeng-Munkoh |
| Wolof | Derguene Mbaye |
| isiXhosa | Andiswa Bukula |
| Yorùbá | Jesujoba Alabi |
| isiZulu | Rooweither Mabuya |
Adding a corpus to the project
It is better to have a folder for each language (folder_name is iso 693-3 letter code) which will have two files,
- corpus with filename (iso 693-3 language code) e.g xho.txt
- A readme file describing the number of articles sentences, and tokens in the corpus. If possible, please specify news categories for the articles, since we prefer a balanced dataset across different categories.