Publications
Research on NLP for low-resource African languages — machine translation, named entity recognition, gender bias, and explainable AI.
2025
Amplify Initiative: Building A Localized Data Platform for Globalized AI
Q. M. Rashid, E. van Liemt, T. Shih, A. Ebinama, K. Barrios Ramos, M. Maji, A. Verma, C. Kalia, J. Smith-Loud, J. Nakatumba-Nabende, R. Baguma, A. Katumba, C. Mutebi, J. Marvin, Eric Wairagala, M. Bruce, P. Oketta, L. Nderu, O. Obiajunwa, A. Oppong, M. Zimba, et al.
Large Consortium · PDF ↗
Collaborative data platform engaging 155 domain experts across five Sub-Saharan African countries to produce 8,091 annotated adversarial queries in seven languages for evaluating AI safety and cultural appropriateness in non-English contexts.
In Progress
Medical Machine Translation for Low-Resourced Languages: A Case Study of Luganda to English
Dr. Joyce Nakatumba-Nabende, Eric Wairagala, Jonathan Mukiibi, Benjamin Akera
In Preparation
Addressing Linguistic Disparities through Explainable and Interpretable AI for Low-Resourced Languages with Large Language Models
Dr. Joyce Nakatumba-Nabende, Eric Wairagala, Jonathan Mukiibi
In Preparation
2022
Gender Bias Evaluation in Luganda–English Machine Translation
Eric Wairagala, Jonathan Mukiibi, Jeremy Francis Tusubira, Claire Babirye, Joyce Nakatumba-Nabende, Andrew Katumba, Ivan Ssenkungu
AMTA 2022 · PDF ↗
First study of gender bias in Luganda–English machine translation; introduces evaluation benchmarks and mitigation strategies.
David I. Adelani, Jesujoba O. Alabi, Angela Fan, Julia Kreutzer, Xiaoyu Shen, Machel Reid, Dana Ruiter, …, Eric Wairagala, et al.
Large Consortium · PDF ↗
Multilingual translation benchmarks spanning 15+ African languages via pre-trained model transfer.
2021
MasakhaNER: Named Entity Recognition for African Languages
David I. Adelani, Jade Abbott, Graham Neubig, Daniel D’Souza, Julia Kreutzer, …, Eric Wairagala, et al.
TACL 2021 · PDF ↗
Foundational NER benchmark across 10 African languages; key resource for low-resource sequence labelling.