Fatima-Bint/african-dishes-knowledge-base

Preserving African indigenous food knowledge and building culturally relevant nutrition tools. # African Dishes Knowledge Base An evidence-first, AI-assisted system for turning fragmented information about African dishes into reviewed, source-backed, searchable records. The application-stage pilot is intentionally limited to Ghana. It demonstrates one complete vertical slice: `source → structured candidate → match suggestion → human review → published record` This is not yet a recipe, calorie, fitness, or medical app. Those layers are deliberately deferred until dish identity and provenance are trustworthy. ## What now works - A Django data model for dishes, alternative names, locations, relationships, sources, evidence excerpts, claims, AI candidates, match suggestions, and review decisions. - A public catalogue that excludes draft and unreviewed records. - Search across canonical and alternative names. - Location and category filters. - Dish pages with claim-level provenance and source links. - JSON API plus JSON and CSV downloads. - A curator admin with auditable human-review actions. - Structured Wikidata ingestion for exact entity IDs. - Optional schema-constrained Gemini extraction from bounded source text. - A deterministic name-matching baseline that runs before approval. - An idempotent 12-record Ghana demo seed and automated tests. ## Quick start Python 3.12+ is recommended. ```bash python -m venv .venv source .venv/bin/activate # Windows: .venv\Scripts\activate pip install -r requirements.txt cp .env.example .env # Windows: copy .env.example .env python manage.py migrate python manage.py seed_demo python manage.py createsuperuser python manage.py runserver ``` Open: - Public catalogue: `127.0.0.1` - Curator review workspace: `127.0.0.1` - Project demo: `127.0.0.1` - JSON API: `127.0.0.1` - JSON export: `127.0.0.1` - CSV export: `127.0.0.1` The seed command creates a non-login audit us …