AfricaNLP Playbook & Annotation Platform → accepted at Deep Learning Indaba 2026.
This session presents two major community-oriented infrastructure projects for African NLP: the AfricaNLP Dataset Creation Playbook and an open, mobile-first annotation platform tailored for low-resource African language contexts. Funded by the Masakhane African Languages Hub, the session is designed to democratize data creation, support grassroots language communities, and strengthen the foundation for African AI research.
Motivation
High-quality, community-owned datasets remain one of the largest bottlenecks in African NLP. While research interest has grown, many communities still lack accessible tools, clear workflows, or governance models for ethical, large-scale, multilingual data creation. Indaba is an ideal venue to expose these tools to the African AI community, gather feedback, and collaboratively shape their next iterations as we approach launch.
Session objectives
- Introduce and demonstrate the AfricaNLP Playbook and annotation platform.
- Hands-on exploration to evaluate usability, accessibility, and relevance for African NLP needs.
- Test real-world annotation tasks across text, speech, and machine translation.
- Create space for community feedback, needs assessment, and collaborative co-design.
- Build bridges between language communities, researchers, practitioners, and tool developers.
- Raise awareness so the wider community is prepared to adopt and deploy these tools.
- Establish a community of interest that continues testing, refining, and contributing.
- Identify missing chapters, feature requests, and long-term collaboration opportunities.
Agenda
A 90-minute showcase and interactive exhibition. Format designed for participatory, hands-on engagement.
- 01 5 min
Opening
Welcome, purpose of the session, and context.
- 02 15 min
Part 1 · Demo of the AfricaNLP Playbook
Overview of chapters (text, speech, MT, governance, sustainability), interactive modules, templates and videos, plus how communities can contribute new chapters or updates.
- 03 15 min
Part 2 · Demo of the Annotation Platform
Mobile-first PWA, offline mode, multilingual support; text, speech and MT annotation workflows; contributor management, dashboards, and quality-control features.
- 04 15 min
Part 3 · Hands-On Exploration
Participants work in small groups: try annotation tasks (voice, text, MT), test the Playbook’s guidelines and templates, try mobile offline mode, and provide live feedback.
- 05 15 min
Part 4 · Community Dialogue
“What is missing for African NLP?” — missing Playbook chapters, needed platform features, African language challenges, and ideas for long-term community involvement.
- 06 5 min
Closing & Next Steps
Summary of insights, invitation to join pilot deployments and long-term testing, and follow-up channels (GitHub, mailing list, social platforms).
Three contributors awarded at the closing
Three team members who actively engaged and showcased real-world examples will receive a recognition award at the end of the Indaba event.
Expected outcomes
Participants leave with concrete tools, knowledge, and connections.
Awareness
Awareness of two major open-source African NLP infrastructure tools.
Practical knowledge
How to use the playbook and platform for your own projects.
Contribution paths
Clear pathways to contribute new chapters, translations, or platform features.
Hands-on experience
Real annotation tasks across text, speech, and machine translation.
Shared understanding
A view of the African community’s needs for data creation.
Live feedback log
A compiled list of feedback, feature requests, and priority use cases.
Target audience
No prior technical background is required.
- African NLP researchers and students
- Linguists and language community leaders
- Annotators and dataset creators
- MT, speech, conversational AI, and text processing practitioners
- NGOs, civic tech groups, and digital rights advocates
- Developers and HCI designers
- Anyone interested in ethical, sustainable African language tech
Organizing team
Members of the two project teams funded by the Masakhane African Languages Hub, anchored at Bayero University Kano (BUK), Bahir Dar University (BDU), and partner institutions.
Seid Muhie Yimam
University of Hamburg · Bahir Dar University
Shamsuddeen Hassan Muhammad
Bayero University Kano
Idris Abdulmumin
University of Pretoria
Abinew Ali Ayele
Bahir Dar University
Tadesse Destaw Belay
IPN
Ibrahim Said Ahmad
University of Wisconsin–Stevens Point