JavaFX RAG Chatbot (Langchain4J + Gemini)

This project is a JavaFX chat application demonstrating Retrieval-Augmented Generation (RAG). It uses Langchain4J to connect to the Google Gemini model and answer questions based on a private knowledge base of PDF documents.

Key Technologies

• Frontend: JavaFX (MaterialFX style).
• LLM: Google Gemini (gemini models).
• RAG Components:
  • Embedding Model: Ollama (embeddinggemma) on http://localhost:11434.
  • Vector Store: ChromaDB on http://localhost:8000.

Prerequisites

1. JDK 21 with JavaFX.
2. Maven.
3. Google Gemini API Key: Must be set in a .env file as GEMINI_API_KEY.
4. ChromaDB and Ollama running locally on their respective default ports.

Setup and Run

1. Prepare Knowledge Base: Place your PDF files in the directory: src/main/resources/knowledge-base/

2. Execute:

   # Build the project
   $ mvn clean install
   
   # Run the application
   $ mvn javafx:run

The RAG pipeline will automatically load and ingest your PDFs into the vector store upon startup.