🎬 Semantic Movie Prompt Generator using MongoDB & Sentence Transformers

This project enriches user-provided screenwriting prompts with semantic context from a movie database.
It leverages sentence-transformers and MongoDB Atlas Vector Search to create intelligent, original story ideas.

---

✨ Features

• Embeds movie plots using sentence-transformers/all-MiniLM-L6-v2
• Stores and retrieves vectors from MongoDB Atlas with $vectorSearch
• Automatically builds detailed prompts for story ideation
• Uses the sample_mflix.movies collection
• Modular scripts for embedding and inference

---

🧠 How It Works

1. Embedding
   The create_embeddings.py script:

   • Connects to MongoDB Atlas
   • Embeds the plot field from each movie
   • Stores the embeddings in a new field (plot_embedding_hf)

2. Inference
   The inference.py script:

   • Takes a user prompt (e.g., "epic medieval war")
   • Finds the top 10 semantically similar movies using $vectorSearch
   • Generates a rich, structured prompt combining those films with the user's idea

---

🔧 Setup

1. Clone and install dependencies

git clone https://github.com/your-repo-name.git
cd your-repo-name
pip install -r requirements.txt

2. Environment Variables

Create a .env file:

USER_NAME=your_mongodb_user
PASSWORD=your_mongodb_password

---

🚀 Running the Embedding Script

python create_embeddings.py

This will iterate over all documents with a plot, generate embeddings, and update them in the MongoDB database.

---

🔍 Running the Prompt Generator

python inference.py

You can modify the query at the bottom of inference.py to test other story ideas.

Example:

query = "A cyberpunk heist in a dystopian city controlled by rogue AIs"

---

📦 Dependencies

Main libraries used:

pymongo
sentence-transformers
python-dotenv

---