MasakhaNER: Named entity recognition for African languages

• Published in: Transactions of the Association for Computational Linguistics
• Main Authors: Adelani, David Ifeoluwa, Abbott, Jade, Neubig, Graham, d'Souza, Daniel, Kreutzer, Julia, Lignos, Constantine, Palen-Michel, Chester, Buzaaba, Happy, Rijhwani, Shruti, Ruder, Sebastian, Mayhew, Stephen, Abebe Azime, Israel, Muhammad, Shamsuddeen H, Chinenye Emezue, Chris, Nakatumba-Nabende, Joyce, Ogayo, Perez, Aremu, Anuoluwapo, Gitau, Catherine, Mbaye, Derguene, Alabi, Jesujoba, Yimam, Seid Muhie, Rabiu Gwadabe, Tajuddeen, Ezeani, Ignatius, Niyongabo, Rubungo Andre, Mukiibi, Jonathan, Otiende, Verrah, Orife, Iroro, David, Davis, Ngom, Samba, Adewumi, Tosin, Rayson, Paul, Adeyemi, Mofetoluwa, Muriuki, Gerald, Anebi, Emmanuel, Chukwuneke, Chiamaka, Odu, Nkiruka, Wairagala, Eric Peter, Oyerinde, Samuel, Siro, Clemencia, Saul Bateesa, Tobius, Oloyede, Temilola, Wambui, Yvonne, Akinode, Victor, Nabagereka, Deborah, Katusiime, Maurice, Awokoya, Ayodele, Mboup, Mouhamadane, Gebreyohannes, Dibora, Tilaye, Henok, Nwaike, Kelechi, Wolde, Degaga, Faye, Abdoulaye, Sibanda, Blessing, Ahia, Orevaoghene, Dossou, Bonaventure F P, Ogueji, Kelechi, Thierno, Ibrahima, Diallo, Abdoulaye, Akinfaderin, Adewale, Marengereke, Tendai, Osei, Salomey
• Format: Journal Article
• Language: English
• Published: The MIT Press 14-06-2021
• Subjects: Computation and Language; Computer Science
• ISSN: 2307-387X
• DOI: 10.1162/tacl

We take a step towards addressing the underrepresentation of the African continent in NLP research by bringing together different stakeholders to create the first large, publicly available, high-quality dataset for named entity recognition (NER) in ten African languages. We detail the characteristics of these languages to help researchers and practitioners better understand the challenges they pose for NER tasks. We analyze our datasets and conduct an extensive empirical evaluation of stateof-the-art methods across both supervised and transfer learning settings. Finally, we release the data, code, and models to inspire future research on African NLP. 1