🎥 Clickbait Clarifier 🚀

Stop wasting time on misleading videos. Clickbait Clarifier uses a hybrid AI approach—combining LightGBM Statistical Learning with Llama 3 Semantic Analysis—to detect and flag deceptive YouTube content in real-time.

---

✨ Key Features

• Hybrid Detection Engine: Uses a LightGBM model trained on 55+ statistical features (engagement ratios, title-to-description consistency, etc.).
• LLM Verification: Deep semantic analysis via Llama 3 (Cerebras Cloud) to verify titles against transcripts.
• Transcript Verification: Fetches and analyzes video transcripts to find "Key Moments" where promises are fulfilled.
• Real-Time Extension: A sleek Chrome extension that adds status badges directly to your YouTube interface.
• Smart Key Rotation: Automatic rotation of Transcripts API keys to handle rate limits and quotas.

---

🏗️ System Architecture

1. High-Level Overview

The system bridges a Chrome content script with a modular Flask backend powered by high-performance AI models.

graph TD
    subgraph Client [Chrome Extension]
        UI[YouTube UI Overlay]
        CS[Content Script]
    end

    subgraph Server [Backend API]
        API[Flask Gateway]
        ML[LightGBM Engine]
        LLM[Llama 3 Brain]
        TR[Transcript Service]
    end

    UI --> CS
    CS -->|POST /predict| API
    API --> ML
    API --> LLM
    TR -->|Context| LLM
    API -->|Verdict| CS

Loading

2. Request Flow (Sequence)

sequenceDiagram
    participant Ext as Chrome Extension
    participant API as Flask Backend
    participant ML as LightGBM
    participant LLM as Llama 3
    
    Ext->>API: POST /predict (video_id)
    par Parallel Analysis
        API->>ML: Statistical Probability
        API->>LLM: Metadata/Transcript Check
    end
    API->>API: Hybrid Ensemble Brain
    API-->>Ext: Verdict & Confidence

Loading

---

🚀 Getting Started

1. Backend Setup (Flask API)

1. Clone the Repository:

   git clone https://github.com/sumedhpatil2005/AntiClickbait.git
   cd AntiClickbait/backend

2. Install Dependencies:

   pip install -r requirements.txt

3. Configure API Keys:

   • Duplicate api_config.example.py to api_config.py.
   • Add your YouTube Data API, Cerebras, and TranscriptAPI.com keys.

4. Run the Server:

   python app.py

2. Extension Setup (Chrome)

1. Open chrome://extensions/ in your browser.
2. Enable "Developer mode" (top right).
3. Click "Load unpacked" and select the /extension folder from this project.
4. Open any YouTube video and look for the detection badge below the title!

---

🛠️ Tech Stack

• Frontend: JavaScript (Chrome Extension API), CSS3 (Glassmorphism UI).
• Backend: Flask (Python), LightGBM, Pandas, Scikit-Learn.
• AI Models: Llama-3 (via Cerebras Cloud), LightGBM (Custom Trained).
• APIs: YouTube Data API v3, TranscriptAPI.com.

---

📜 Acknowledgments

• Cerebras Cloud for the blazing-fast Llama 3 inference.
• TranscriptAPI for the robust YouTube subtitle retrieval.

---

Created with ❤️ for a safer, cleaner YouTube experience.