MasakhaNER 2.0: Africa-centric Transfer Learning for Named Entity Recognition

African languages are spoken by over a billion people, but are underrepresented in NLP research and development. The challenges impeding progress include the limited availability of annotated datasets, as well as a lack of understanding of the settings where current methods are effective. In this pa...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2022-11
Main Authors: Adelani, David Ifeoluwa, Neubig, Graham, Ruder, Sebastian, Rijhwani, Shruti, Beukman, Michael, Palen-Michel, Chester, Lignos, Constantine, Alabi, Jesujoba O, Muhammad, Shamsuddeen H, Nabende, Peter, Cheikh M Bamba Dione, Bukula, Andiswa, Mabuya, Rooweither, Bonaventure F P Dossou, Sibanda, Blessing, Buzaaba, Happy, Mukiibi, Jonathan, Godson Kalipe, Mbaye, Derguene, Taylor, Amelia, Kabore, Fatoumata, Emezue, Chris Chinenye, Aremu, Anuoluwapo, Perez Ogayo, Gitau, Catherine, Munkoh-Buabeng, Edwin, Koagne, Victoire M, Allahsera Auguste Tapo, Macucwa, Tebogo, Marivate, Vukosi, Mboning, Elvis, Gwadabe, Tajuddeen, Adewumi, Tosin, Ahia, Orevaoghene, Joyce Nakatumba-Nabende, Mokono, Neo L, Ezeani, Ignatius, Chukwuneke, Chiamaka, Adeyemi, Mofetoluwa, Hacheme, Gilles Q, Idris Abdulmumin, Ogundepo, Odunayo, Yousuf, Oreen, Ngoli, Tatiana Moteu, Klakow, Dietrich
Format: Article
Language:English
Subjects:
Datasets
Languages
R&D
Recognition
Research & development
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:African languages are spoken by over a billion people, but are underrepresented in NLP research and development. The challenges impeding progress include the limited availability of annotated datasets, as well as a lack of understanding of the settings where current methods are effective. In this paper, we make progress towards solutions for these challenges, focusing on the task of named entity recognition (NER). We create the largest human-annotated NER dataset for 20 African languages, and we study the behavior of state-of-the-art cross-lingual transfer methods in an Africa-centric setting, demonstrating that the choice of source language significantly affects performance. We show that choosing the best transfer language improves zero-shot F1 scores by an average of 14 points across 20 languages compared to using English. Our results highlight the need for benchmark datasets and models that cover typologically-diverse African languages.
ISSN:2331-8422