How mohammedmahboob786/movierag works
Project type and purpose
• This is a retrieval-augmented movie identification application (“MovieRAG”) that ingests audio clips, converts them to text, and uses vector-based similarity search over a prebuilt movie/TV series corpus to identify the source.
• It is organized as a three-stage pipeline: Data Preprocessing, Indexing, and Application (the RAG service exposed via Streamlit).
File structure
• Top level:
• Data_Preprocessing)Project_4_Db_retireval.ipynb – notebook for raw data cleaning/extraction
• Indexing)Project_4_Db_retrieval.ipynb – notebook for splitting text, generating embeddings, and persisting them
• RAG_app(using gte-small embeddings).py – the Streamlit-based application code
• README.md – high-level overview and workflow
• Absence of separate frontend/backend folders indicates a single-service Python deployment (Streamlit as both UI and orchestration).
README for additional insights
• Describes three phases (preprocessing, indexing, application).
• Lists core technologies: Whisper for ASR, BeautifulSoup & regex for cleaning, LangChain for loaders/chunkers/chains, Hugging Face BERT (“thenlper/gte-small”) for embeddings, ChromaDB for vector storage, and Streamlit for UI.