Resources & Tools

Below is a collection of datasets, tools, software, and guides that I have developed or contributed to, all publicly available for the research community.

Quick Navigation

Jump to any section:

• Datasets — Research datasets for African languages
• Tools & Software — NLP tools and platforms
• Writing NLP Papers — Guides and templates
• Other Resources — Tutorials, and more

Datasets

BRIGHTER: Multilingual Emotion Recognition Dataset

• Description: Human-annotated textual emotion recognition datasets for 28 languages
• Languages: 28 languages including African, Asian, and European languages
• Task: Emotion classification
• Paper: ACL 2025 (Best Resource Paper Award)
• Access: Website • HuggingFace

AfriSenti: African Sentiment Analysis Dataset

• Description: Twitter sentiment analysis benchmark for African languages
• Languages: 14 African languages (Hausa, Yoruba, Igbo, Amharic, Swahili, and more)
• Task: Sentiment classification (positive, negative, neutral)
• Paper: EMNLP 2023 (Best Non-archival Paper Award)
• Access: GitHub • HuggingFace • Leaderboard

NaijaSenti: Nigerian Sentiment Analysis Dataset

• Description: Sentiment analysis dataset specifically for Nigerian languages and Nigerian Pidgin
• Languages: Hausa, Yoruba, Igbo, Nigerian Pidgin
• Task: Sentiment classification (positive, negative, neutral)
• Paper: AfricaNLP Workshop
• Access: GitHub • HuggingFace

AfriHate: Hate Speech & Abusive Language Dataset

• Description: Multilingual collection of hate speech and abusive language datasets for African languages
• Recognition: IRCAI Global Top 100 AI Projects Award 2025
• Languages: Multiple African languages
• Task: Hate speech detection, abusive language classification
• Paper: NAACL 2025
• Access: GitHub • HuggingFace

MasakhaNER: Named Entity Recognition Dataset

• Description: Named entity recognition datasets for 21 African languages
• Languages: Amharic, Hausa, Igbo, Kinyarwanda, Luganda, Luo, Nigerian-Pidgin, Swahili, Wolof, Yoruba, and more
• Task: Named entity recognition (PER, LOC, ORG, DATE)
• Paper: TACL 2021
• Access: GitHub • HuggingFace

BLEnD: Everyday Knowledge Dataset

• Description: Everyday knowledge in diverse cultures and languages for evaluating cultural understanding in LLMs
• Languages: Multiple languages representing diverse cultures
• Task: Cultural knowledge understanding and reasoning
• Paper: NeurIPS 2024 (Best Non-archival Paper Award)
• Access: GitHub • HuggingFace

SemRel: Semantic Textual Relatedness Dataset

• Description: Semantic textual relatedness datasets for African and Asian languages
• Languages: 14 languages from Africa and Asia
• Task: Semantic similarity and relatedness measurement
• Paper: SemEval 2024 (Honourable Mention)
• Access: GitHub • arXiv

AFRIDOC-MT: African Document-level Machine Translation

• Description: Document-level machine translation dataset for African languages with document context
• Languages: Multiple African languages
• Task: Document-level machine translation
• Paper: EMNLP 2025
• Access: HuggingFace

INJONGO: Multimodal Dataset for African Languages

• Description: Multimodal benchmark evaluating vision-language models on African cultural contexts
• Languages: Multiple African languages
• Task: Vision-language understanding and reasoning
• Paper: ACL 2025
• Access: GitHub

POLAR: Persuasive & Offensive Language in African Regional Languages

• Description: Dataset for detecting persuasive and offensive language in African regional contexts
• Languages: African regional languages
• Task: Persuasive language detection, offensive language identification
• Paper: arXiv
• Access: GitHub

UHURA: Unified Dataset for African Languages

• Description: Comprehensive unified dataset for multiple NLP tasks in African languages
• Languages: Multiple African languages
• Tasks: Multiple NLP tasks
• Paper: arXiv
• Access: GitHub

AfroXLMR-Social: African Social Media Dataset

• Description: Social media dataset for African languages covering diverse social contexts
• Languages: Multiple African languages
• Task: Social media text processing and understanding
• Access: HuggingFace

Who Wrote This: Hausa Authorship Attribution Dataset

• Description: Dataset for authorship attribution and text classification in Hausa
• Language: Hausa
• Task: Authorship attribution, text classification
• Paper: arXiv
• Access: GitHub

HausaHate: Hausa Hate Speech Dataset

• Description: Hate speech detection dataset specifically for Hausa language
• Language: Hausa
• Task: Hate speech detection and classification
• Paper: WOAH 2024
• Access: GitHub

AfriGSM: African Grade School Math Dataset

• Description: Grade school mathematics reasoning dataset for African languages
• Languages: Multiple African languages
• Task: Mathematical reasoning and problem solving
• Access: HuggingFace

AfriMMLU: African Massive Multitask Language Understanding

• Description: Massive multitask language understanding benchmark for African languages
• Languages: Multiple African languages
• Tasks: Multiple choice question answering across various domains
• Access: HuggingFace

AfriXNLI: African Cross-lingual Natural Language Inference

• Description: Cross-lingual natural language inference dataset for African languages
• Languages: Multiple African languages
• Task: Natural language inference (entailment, contradiction, neutral)
• Access: HuggingFace

AfriCOMET: African Machine Translation Evaluation Metric

• Description: Evaluation metric specifically designed for assessing machine translation quality in African languages
• Languages: Multiple African languages
• Task: Machine translation evaluation
• Access: GitHub

MasakhaNEWS: African Language News Classification

• Description: News topic classification dataset for African languages
• Languages: Multiple African languages
• Task: News topic classification
• Access: HuggingFace

MasakhaPOS: African Language Part-of-Speech Tagging

• Description: Part-of-speech tagging dataset for African languages
• Languages: Multiple African languages
• Task: Part-of-speech tagging
• Access: HuggingFace

HaVQA: Hausa Visual Question Answering

• Description: Visual question answering dataset for Hausa language
• Language: Hausa
• Task: Visual question answering
• Access: GitHub

Hausa Visual Genome

• Description: Visual genome dataset with Hausa language annotations for image understanding
• Language: Hausa
• Task: Image captioning, visual understanding
• Access: HuggingFace

HERDPhobia: Hate Speech and Extreme Religious Discourse Dataset

• Description: Dataset focusing on hate speech and extreme religious discourse in Hausa
• Language: Hausa
• Task: Hate speech detection, religious discourse analysis
• Access: GitHub

MAFAND-MT: Machine Translation for African Languages

• Description: Machine translation dataset for African languages
• Languages: Multiple African languages
• Task: Machine translation
• Access: GitHub

FLORES Fix for Africa

• Description: Corrected and improved version of FLORES evaluation dataset for African languages
• Languages: African languages in FLORES
• Task: Machine translation evaluation
• Access: GitHub

Nigerian Speech Corpus

• Description: Large-scale speech corpus for major Nigerian languages
• Funding: Lacuna Fund ($120,000)
• Languages: Hausa, Yoruba, Igbo, and Nigerian Pidgin
• Tasks: Automatic speech recognition, text-to-speech synthesis
• Status: In development
• Access: Project Website

Tools & Software

Hausa Dictionary (Kamusun Hausa)

• Description: An online lexicon providing standardized Hausa definitions, examples, and usage
• Features: Comprehensive Hausa-English dictionary, example sentences, pronunciation guide, standardized definitions
• Access: https://kamusunhausa.hausanlp.org/

Hausa Catalogue

• Description: A centralized repository indexing datasets, benchmarks, and resources for Hausa NLP research
• Features: Searchable database of datasets, research papers, models, and tools for Hausa language
• Access: https://catalog.hausanlp.org

Annotate Platform

• Description: A lightweight web-based annotation platform for text classification, translation, and sequence labeling
• Features: Multi-task annotation support, collaborative annotation, quality control, export to standard formats
• Use Cases: Text classification, machine translation, named entity recognition, sentiment labeling
• Access: https://annotate.hausanlp.org

NLPQuiz

• Description: Interactive quiz platform for learning and testing NLP concepts
• Features: Educational quizzes, progress tracking, concept reinforcement
• Access: https://nlpquiz.hausanlp.org/auth/login

Speech Annotation Bot

• Description: A Telegram-based tool for collecting speech data in low-connectivity environments; optimized for mobile users
• Features: Mobile-first design, works in low-bandwidth areas, Telegram integration, easy audio collection
• Use Cases: Speech corpus collection, crowdsourced pronunciation data, dialect documentation
• Platform: Telegram Bot
• Languages: Hausa and other low-resource languages

Writing NLP Papers

Paper Templates & Examples

• ACL Conference Templates:
  • ACL 2025 LaTeX Template (Overleaf)
  • NAACL Paper Template
  • EMNLP Submission Template
• Example Papers from My Work:
  • Dataset Paper Example — BRIGHTER (ACL 2025)
  • Shared Task Paper Example — SemEval Task 11
  • Benchmark Paper Example — IrokoBench (NAACL 2025)
  • System Description Example — AfriSenti Shared Task

Writing Guides & Best Practices

• Essential Reading:
  • ACL Author Guidelines
  • How to Write a Strong NLP Paper
  • Effective Abstract Writing for ML/NLP
  • Results & Analysis Section Guide
  • Common Pitfalls to Avoid in NLP Papers
• My Personal Recommendations:
  • Start with a clear research question and contribution statement
  • Use active voice and concise language
  • Include comprehensive error analysis and ablation studies
  • Make your code and data publicly available
  • Get feedback early and often from colleagues
  • Address limitations and ethical considerations explicitly

LaTeX Tips & Resources

• Useful LaTeX Packages:
  • booktabs — Professional-looking tables
  • algorithm2e — Algorithm formatting
  • tikz — Diagrams and figures
  • natbib or biblatex — Bibliography management
  • hyperref — Cross-references and URLs
• Helpful Tools:
  • Overleaf — Online collaborative LaTeX editor
  • Tables Generator — Quick table creation
  • Detexify — Find LaTeX symbols by drawing

Understanding the Peer Review Process

• Typical Timeline:
  • Submission → Review assignment (2-3 weeks)
  • Review period (6-8 weeks)
  • Author response period (1 week)
  • Final decision (2-4 weeks after response)
• Tips for Strong Author Responses:
  • Be respectful and professional in all responses
  • Address each reviewer concern point-by-point
  • Provide specific line numbers for changes made
  • Acknowledge valid criticisms graciously
  • Clarify misunderstandings politely with evidence
• Additional Resources:
  • How to Write an Effective Author Response
  • Dealing with Paper Rejection Constructively
  • Ethics in NLP Research
  • ACL Code of Ethics

Tipsa and Resources

Tutorials & Workshop Materials

• Video Tutorials:

  Community & Collaboration

• Research Organizations:
  • HausaNLP — Hausa language NLP research group
  • Arewa Data Science Academy — Free AI/ML training for underserved students
  • Masakhane — Grassroots African NLP community
• Tips for Successful Grant Applications:
  • Clearly articulate the problem and potential impact
  • Demonstrate strong community involvement and partnerships
  • Include realistic budgets with detailed justifications
  • Highlight sustainability and scalability plans
  • Show evidence of preliminary work or pilot studies

Acknowledgments

These resources were made possible through awosome collaborators and generous funding support from:

• Google (DeepMind Academic Fellowship)
• Lacuna Fund
• Oracle Cloud Infrastructure (OCI)
• Nigerian TETFund
• Wikimedia Foundation
• University of Porto
• Imperial College London
• And countless volunteers and community members who contributed their time and expertise

Last updated: December 2025