AI Social Search

A Chrome extension that uses AI to improve YouTube search results with two powerful modes:

• Smart Reranker: Automatically reorders search results based on semantic relevance
• Tag Collector: Finds and collects videos matching specific tags across hundreds of results

Quick Start Guide

Step 1: Install Python Backend

1. Install Python (if you don't have it):

   • Download Python 3.10 or newer from python.org
   • During installation, check "Add Python to PATH"

2. Open PowerShell in the project folder:

   • Right-click the folder → "Open in Terminal" or "Open PowerShell window here"

3. Create a virtual environment:

   python -m venv .venv

4. Activate the virtual environment:

   .venv\Scripts\Activate.ps1

5. Install dependencies:

   pip install -r backend/requirements.txt

   ⏱️ This may take a few minutes. First run will download AI models (~500MB).

6. Start the backend:

   python backend/app.py

   You should see: Running on http://127.0.0.1:5000

   Keep this window open while using the extension.

Step 2: Install Chrome Extension

1. Open Google Chrome and go to: chrome://extensions

2. Toggle Developer mode ON (top-right corner)

3. Click "Load unpacked"

4. Select the extension folder from this project

5. You should see "AI Social Search" appear in your extensions

Step 3: Use the Extension

For Smart Reranker Mode:

1. Go to YouTube and search for anything (e.g., "apple")
2. Click the extension icon in your browser toolbar
3. Click "Smart Reranker" mode
4. Click "Save" to enable
5. Refresh the YouTube page
6. Results are now reordered by AI relevance with colored highlights!

For Tag Collector Mode:

1. Go to YouTube and search broadly (e.g., "art")
2. Click the extension icon
3. Click "Tag Collector" mode
4. Select tags or type custom ones (e.g., "watercolor, painting")
5. Add negative tags to exclude (e.g., "music, shorts")
6. Set min score (15-30 recommended) and max videos (50-100)
7. Click "Start Collecting"
8. The page will auto-scroll and collect matching videos
9. Copy links or create a playlist when done!

Troubleshooting

Backend won't start?

• Make sure you activated the virtual environment (.venv\Scripts\Activate.ps1)
• Try: pip install --upgrade pip then reinstall requirements

Extension not working?

• Make sure the backend is running (check PowerShell window)
• In the extension popup, verify the URL is http://127.0.0.1:5000/search
• Check that "Enable semantic reranking" is ON

No videos collected?

• Lower the minimum score (try 15%)
• Make sure you're on a YouTube search results page
• Try broader search terms first

Features

Smart Reranker

• AI-powered semantic search reranking
• Visual thumbnail matching using CLIP
• Filters out irrelevant music/entertainment content
• Smart query intent detection (tutorials, reviews, factual content)
• Temporal boost for recent/trending videos

Tag Collector

• Auto-scroll through hundreds of YouTube results
• Match videos by semantic tags
• Exclude unwanted content with negative tags
• User feedback blocklist (thumbs down to never see again)
• Export to playlist or copy all links

How It Works

The backend uses AI models (SentenceTransformers + CLIP) to understand the meaning of your search and video content, not just keywords. This means:

• Searching "apple" shows cooking videos, not iPhone reviews
• Videos are ranked by actual relevance, not just view count
• Thumbnails are checked to match visual expectations

Project Structure

├── backend/           # Python Flask API with AI models
│   ├── app.py        # Main backend server
│   └── requirements.txt
├── extension/         # Chrome extension
│   ├── manifest.json # Extension config
│   ├── background.js # Handles API calls
│   ├── content.js    # Modifies YouTube page
│   └── popup.html    # Extension settings UI
└── README.md

Team members

• Jennifer Corte
• Eishah Shah
• Jeraldine Rodriguez