Recipe-Search

Recipe-Search is the backend service for the Recipe Social website. It is built with Flask and integrates a custom trained machine learning model and natural language processing techniques to search and retrieve recipes from the internet.

This backend is designed to:

• Use an ML model (trained using XGBoost) to intelligently scrape recipe content from various websites.
• Apply NLP (with spaCy) to enhance the extracted recipe data.
• Store the processed recipes in a Supabase database for easy access.
• Allow users to search for certain recipes using Natural Language Processing techniques.

---

Project Overview

The Flask server provides API endpoints that interact with a trained ML model and NLP pipeline to:

• Scrape recipe information from web pages.
• Parse and enhance the scraped content using natural language processing.
• Save the finalised recipe data into a Supabase Postgresql database.
• Allow for easy query-based recipe matching to easily find recipes

---

Machine Learning and NLP

• ML Model:

  • Algorithm: XGBoost
  • Training: Custom-trained to detect valid recipe pages based on web page content.
  • Role: To decide whether a page contains a recipe worth extracting.

• Natural Language Processing:

  • Library: spaCy
  • Purpose: To process and clean the text, extract key entities (like ingredients and steps), and allow for easy query-based recipe matching.

---

Technologies Used

• Flask — Lightweight backend framework for serving API endpoints.
• Python — Core programming language.
• XGBoost — ML model training and predictions.
• spaCy — NLP for text processing and enhancement.
• BeautifulSoup / Requests — Web scraping and parsing libraries.
• Supabase — Database storage and management.

---

Key Features

• Smart Scraping:
  Only web pages with verified, model-approved recipes are scraped and stored.

• NLP-Enhanced Structuring and Matching:
  Recipes are properly matched with queries from the frontend or the Node-based backend using NLP techniques to produce accurate results.

• Automated Storage:
  Successfully extracted recipes are automatically inserted into the Supabase database.

• Extensible Design:
  The backend is structured to allow easy model updates, additional scraping targets, and future feature expansion.

---

Contact

For any queries, suggestions, or improvements, feel free to open a discussion or reach out!