A5089451523- Natural Language Processing Techniques
- Topic Modeling
- Text Readability and Simplification
- Environmental Education and Sustainability
- Transboundary Water Resource Management
- ICT in Developing Communities
- Multimodal Machine Learning Applications
- ICT Impact and Policies
- Misinformation and Its Impacts
- Financial Risk and Volatility Modeling
- International Development and Aid
- AI in Service Interactions
- Political Conflict and Governance
- Climate variability and models
- Private Equity and Venture Capital
- Religion, Ecology, and Ethics
- Climate Change Communication and Perception
- Semantic Web and Ontologies
- Cooperative Studies and Economics
- Hydrology and Drought Analysis
- Digital Economy and Work Transformation
- African history and culture analysis
- Economic and Technological Innovation
- Multilingual Education and Policy
- Politics and Conflicts in Afghanistan, Pakistan, and Middle East
Microsoft (United States)
2024
Microsoft (Finland)
2023
Johns Hopkins University
2023
Kabir Ahuja, Harshita Diddee, Rishav Hada, Millicent Ochieng, Krithika Ramesh, Prachi Jain, Akshay Nambi, Tanuja Ganu, Sameer Segal, Mohamed Ahmed, Kalika Bali, Sunayana Sitaram. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.
David Adelani, Jesujoba Alabi, Angela Fan, Julia Kreutzer, Xiaoyu Shen, Machel Reid, Dana Ruiter, Dietrich Klakow, Peter Nabende, Ernie Chang, Tajuddeen Gwadabe, Freshia Sackey, Bonaventure F. P. Dossou, Chris Emezue, Colin Leong, Michael Beukman, Shamsuddeen Muhammad, Guyo Jarso, Oreen Yousuf, Andre Niyongabo Rubungo, Gilles Hacheme, Eric Wairagala, Muhammad Umair Nasir, Benjamin Ajibade, Tunde Ajayi, Yvonne Gitau, Jade Abbott, Mohamed Ahmed, Millicent Ochieng, Anuoluwapo Aremu, Perez...
Generative AI models have shown impressive performance on many Natural Language Processing tasks such as language understanding, reasoning, and generation. An important question being asked by the community today is about capabilities limits of these models, it clear that evaluating generative very challenging. Most studies LLMs been restricted to English unclear how capable are at understanding generating text in other languages. We present first comprehensive benchmarking - MEGA, which...
Despite the widespread adoption of Large language models (LLMs), their remarkable capabilities remain limited to a few high-resource languages. Additionally, many low-resource languages (e.g. African languages) are often evaluated only on basic text classification tasks due lack appropriate or comprehensive benchmarks outside In this paper, we introduce IrokoBench -- human-translated benchmark dataset for 16 typologically-diverse covering three tasks: natural inference~(AfriXNLI),...
The deployment of Large Language Models (LLMs) in real-world applications presents both opportunities and challenges, particularly multilingual code-mixed communication settings. This research evaluates the performance seven leading LLMs sentiment analysis on a dataset derived from WhatsApp chats, including Swahili, English Sheng. Our evaluation includes quantitative using metrics like F1 score qualitative assessment LLMs' explanations for their predictions. We find that, while Mistral-7b...
This white paper is the output of a multidisciplinary workshop in Nairobi (Nov 2023). Led by cross-organisational team including Microsoft Research, NEPAD, Lelapa AI, and University Oxford. The brought together diverse thought-leaders from various sectors backgrounds to discuss implications Generative AI for future work Africa. Discussions centred around four key themes: Macroeconomic Impacts; Jobs, Skills Labour Markets; Workers' Perspectives Africa-Centris Platforms. provides an overview...
Recent advances in the pre-training of language models leverage large-scale datasets to create multilingual models. However, low-resource languages are mostly left out these datasets. This is primarily because many widely spoken not well represented on web and therefore excluded from crawls used Furthermore, downstream users restricted selection originally chosen for pre-training. work investigates how optimally existing pre-trained translation systems 16 African languages. We focus two...
In the Global South, effects of climate change have resulted in more frequent and severe weather events such as droughts, floods, storms, leading to crop failures, food insecurity, job loss. These are expected increase intensity future, further disadvantaging already marginalized communities exacerbating existing inequalities. Hence, need for prevention adaptation is urgent, but accurate forecasting remains challenging, despite advances machine learning numerical modeling, due complex...
There has been a surge in LLM evaluation research to understand capabilities and limitations. However, much of this confined English, leaving building for non-English languages relatively unexplored. Several new LLMs have introduced recently, necessitating their on languages. This study aims perform thorough the SoTA (GPT-3.5-Turbo, GPT-4, PaLM2, Gemini-Pro, Mistral, Llama2, Gemma) by comparing them same set multilingual datasets. Our benchmark comprises 22 datasets covering 83 languages,...
Despite the recent progress on scaling multilingual machine translation (MT) to several under-resourced African languages, accurately measuring this remains challenging, since evaluation is often performed n-gram matching metrics such as BLEU, which typically show a weaker correlation with human judgments. Learned COMET have higher correlation; however, lack of data ratings for complexity annotation guidelines like Multidimensional Quality Metrics (MQM), and limited language coverage...
Mobile messaging apps and SMS-based tools have been deployed to extend healthcare services beyond the clinic; peer support chat groups, consisting of patients providers, can improve medication adherence. However, moderation be burdensome for busy professionals who must respond patients, provide accurate timely information, engage build community among patients. In this paper, taking an ethnographic approach, we examine groups young people living with HIV in Kenya. We describe roles...