🔍 RAG Historical Events Explorer

This is a Retrieval-Augmented Generation (RAG) system built to answer questions about 20th-century historical events. It features a sleek, responsive frontend built in Rust (Yew/WebAssembly) and a powerful backend built in Python (Flask).

The system retrieves relevant documents from a FAISS vector index (created from DBpedia data) and uses them as context for a Large Language Model (either local Ollama or OpenAI) to generate an accurate, grounded answer.

✨ Features

• Dual Backend Support: Seamlessly toggle between a local Ollama model (phi3:mini) or the remote OpenAI API (gpt-4-turbo).

• Adjustable Context: Use a slider to control the number of retrieved documents (top_k) to be fed to the model.

• Semantic Search: Uses sentence-transformers and FAISS to perform fast, accurate semantic search over the historical event database.

• Rich Data Source: The knowledge base is built from 20th-century historical events pulled from DBpedia using SPARQL.

• Modern Rust Frontend: A fast, fully client-side application built with Yew, which compiles to WebAssembly.

🛠️ Tech Stack

Component	Technology	Description
Frontend	Rust + Yew	A modern framework for building multi-threaded client-side web apps with WebAssembly.
Frontend	Trunk	Build tool for Yew applications.
Frontend	SASS/SCSS	For advanced and clean styling.
Backend	Python + Flask	A lightweight web server to host the RAG API.
RAG Pipeline	Ollama (phi3:mini)	The default, local LLM for generation.
RAG Pipeline	OpenAI (gpt-4-turbo)	The optional, high-performance LLM.
RAG Pipeline	FAISS	A library for efficient similarity search in vector databases.
RAG Pipeline	sentence-transformers	For generating high-quality text embeddings (all-MiniLM-L6-v2).

🚀 Setup and Installation

Follow these steps to get the complete application running on your local machine.

1. Prerequisites

Before you begin, ensure you have the following tools installed:

• Python 3.10+

• Rust & Cargo

• Trunk: The build tool for Yew. Install it via Cargo:

    cargo install trunk

• Ollama: Download and install the desktop application.

2. Backend Setup

This will set up the Python server, download the ML models, and gather the data.

1.Clone the repository (if you haven't):

git clone https://github.com/frontinus/Generative_IA_Wiki.git
cd Generative_IA_Wiki

2.Create a virtual environment and install dependencies:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

3.Download the Ollama LLM: Pull the phi3:mini model that the pipeline uses by default.

ollama pull phi3:mini

4.Gather Data from DBpedia:

Run the data gathering script. This will create the historical_events_with_abstracts.csv file.

python data_gathering.py

5.(Optional) Set OpenAI API Key:

If you want to use the OpenAI toggle, set your API key as an environment variable.

export OPENAI_API_KEY="sk-YourActualApiKeyHere"

3. Frontend Setup

The frontend is in a separate directory (small_interface) and requires no additional setup, as trunk will handle all dependencies.

▶️ How to Run

You will need three separate terminals running at the same time.

Terminal 1: Run Ollama

Ensure the Ollama Desktop application is running. Alternatively, you can run the service from your terminal:

ollama serve

Terminal 2: Run the Flask Backend (API)

This terminal runs the Python server, which will load the FAISS index into memory and serve the RAG API.

# Navigate to the project root
cd Generative_IA_Wiki

# Activate your virtual environment
source venv/bin/activate

# Run the app
python app.py

Your backend is now running at http://127.0.0.1:8000.

Terminal 3: Run the Yew Frontend (UI)

This terminal builds and serves the Rust-based user interface.

# Navigate to the frontend directory
cd small_interface

# Serve the app. Trunk will compile and open your browser.
trunk serve --open

Your frontend is now running at http://127.0.0.1:8080/. You can now interact with the application!

📁 Project Structure

.
├── app.py                  # Flask backend server
├── pipeline.py             # RAG logic (FAISS, Ollama, OpenAI)
├── data_gathering.py       # SPARQL script to get data from DBpedia
├── requirements.txt        # (Assumed) Python dependencies
├── historical_events_with_abstracts.csv  # (Generated by data_gathering.py)
│
├── small_interface/        # Yew frontend folder
│   ├── src/
│   │   └── main.rs         # Yew app logic
│   ├── index.html          # Entrypoint for frontend
│   └── styles.scss         # App styling
│
└── README.md               # This file

👨‍💻 Authors

This project was built by:

• Francesco
  • GitHub
  • LinkedIn
• Thomas
  • GitHub
  • LinkedIn