Semantic Food Recipe Finder

Semantic Food Recipe Finder is a full-stack semantic web project that builds a recipe ontology, generates an RDF knowledge graph from structured datasets, and exposes a queryable API with a lightweight front-end. Users can explore the knowledge graph with semantic filters such as “vegan lentil recipes under 45 minutes.”

Features

• Ontology-driven graph – Scripts for transforming the Kaggle “Better Recipes for a Better Life” dataset into a Turtle knowledge graph using rdflib with inferred diet categories.
• Semantic search API – A Flask backend that performs SPARQL queries over the graph, supporting ingredient keyword searches, cuisine filtering, diet constraints, and total-time limits.
• Interactive front-end – A responsive interface that consumes the API, visualises recipe metadata, and displays full directions straight from the graph.
• Extensible ontology – Classes for recipes, ingredients, cuisines, and dietary types that can be customised for coursework or further research.

Project Structure

.
├── app/                  # Flask application and graph helpers
├── data/                 # Sample data and RDF graph
├── frontend/             # Static front-end assets
├── scripts/              # Utilities for building the graph
├── requirements.txt      # Python dependencies
└── README.md

Key Files

• scripts/build_graph.py – Convert the Kaggle dataset (CSV or JSON) into an RDF Turtle file with inferred dietary categories.
• data/recipes_sample.ttl – A small demonstration graph generated from the sample JSON dataset.
• app/main.py – Flask entry-point that exposes the API and serves the static site.
• app/graph_loader.py – High-level search helpers that wrap SPARQL queries and assemble JSON-friendly objects.
• frontend/index.html, styles.css, app.js – Search UI that calls the API and renders recipe details.

Quick Start (Recommended)

The easiest way to get started:

Option 1: Automatic Startup (Recommended)

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run the automatic startup script
./start.sh    # On Mac/Linux
# OR
start.bat     # On Windows

Option 2: Manual Startup

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run the application (automatically finds free port)
python run.py

The application will automatically:

• Find a free port (8000, 8001, 8002, etc.)
• Load your recipe database
• Show the URL to open in your browser
• Display recipes immediately when you open the page

Detailed Setup

1. Set up Python dependencies

Create a virtual environment and install requirements (Python 3.10+ recommended):

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt

2. Run with sample data (easiest)

The project includes sample data that works out of the box:

python run.py

This will:

• Automatically build the sample knowledge graph if needed
• Start the Flask server on http://localhost:5000
• Show you helpful startup information

3. Run with your own dataset (optional)

If you want to use your own recipe data:

1. Download the Kaggle dataset:

   • Visit: https://www.kaggle.com/datasets/thedevastator/better-recipes-for-a-better-life
   • Download the CSV file and save it as data/recipes.csv

2. Build your knowledge graph:

   python scripts/build_graph.py data/recipes.csv data/recipes.ttl

3. Set the environment variable:

   export RECIPE_GRAPH_PATH=$(pwd)/data/recipes.ttl

4. Run the application:

   python run.py

4. Alternative: Manual Flask startup

If you prefer to use Flask directly:

export FLASK_APP=app.main:create_app
flask run

API Endpoints

• GET /api/search – Query recipes with optional ingredient, cuisine, diet, and maxTime parameters
• GET /api/recipes?uri=<URI> – Retrieve full details for a recipe resource
• GET /api/cuisines and GET /api/diets – Retrieve available filter values
• GET /api/health – Health check endpoint

Visit http://localhost:5000 to load the front-end UI.

Front-end Overview

The single-page interface provides:

• A search form with ingredient keyword, cuisine dropdown, diet dropdown, and total time filter.
• Live query results that display ratings, cuisine, diets, and cook time badges.
• A details panel that renders the recipe’s ingredients and directions fetched directly from the knowledge graph.

Extending the Ontology

The default ontology defines the following key classes and properties:

• rec:Recipe, rec:Ingredient, rec:Cuisine, rec:DietType
• Properties: rec:hasIngredient, rec:hasCuisine, rec:hasDiet, rec:hasTotalTime, rec:hasRating, and more
• Ingredient subclasses for rec:AnimalProduct and rec:GlutenIngredient

To extend the ontology, edit scripts/build_graph.py to add new classes or properties, then regenerate the TTL file. The Flask API automatically reflects additional data properties because SPARQL queries use labels and groups.