Overview

Ever find yourself hungry, aimlessly scrolling through Instagram, trying to track down that one magical restaurant the algorithm teased you with earlier? Same here. Even when I do save posts, it's a hassle to find them later, look up the restaurant's location, and figure out what they serve. So, with an hour to kill and my local llama3 model ready to play, I thought, why not automate the whole process?

I will most likely not continue this project, but I thought it would be fun to share

Dataset:

HuggingFace Dataset

Usage

Step 1: Scrape Instagram Captions

Run the caption_scraper.ipynb notebook to fetch captions from the specified Instagram profiles and save them to JSON files.

Step 2: Extract Restaurant Information

Run the restaurant_extractor.ipynb notebook to extract detailed restaurant information from the saved captions and save the results.

Step 3: Post-Process Data

Run the postpro.ipynb notebook to clean and filter the extracted restaurant data, ensuring it is ready for use.

Step 4: Spin Up the Web App

Run the app.py file to start the Flask server and access the restaurant recommendations via the browser.

---

Work In Progress:

• Generate the restaurant recommendations for some of the pages not already processed
• Make the service text-based/slackbot based?