Get started

Back

cd ./back/
pip install requirement.txt
python -m uvicorn src.main:app --reload

Front

cd ./front/
npm install
ng serve

Documentation

You can see the Swagger after starting the project at http://localhost:8000/docs

How does it work ?

Update

1. Get all the articles from the RSS flux
2. Get the embeding of the title from google's api
3. Save the embeding and the article in a vector database faiss

Search

1. Get the embeding of the query from google's api
2. Get the 3 closest vector from the database
3. Return there metadata

API Key Setup

To run the backend, you need to add your API key to the environment configuration file:

1. Open the file:

   back/src/env.py
   

2. Add your API key in the following format:

   GEMINI_API_KEY = "YOUR_GOOGLE_API_KEY"

3. You can create and manage your API key at: https://aistudio.google.com/app/apikey

Ressources

FAISS github
Obtenir des embedding de texte
Vectoring Words (Word Embeddings), Computerphile
Camembert model
Text embeddings & semantic search, HuggingFace

Feedparser documentation
RSS made easy, Growth origin
Public flux rss
VSD flux rss

Reference

@article{douze2024faiss,
title={The Faiss library},
author={Matthijs Douze and Alexandr Guzhva and Chengqi Deng and Jeff Johnson and Gergely Szilvasy and Pierre-Emmanuel Mazaré and Maria Lomeli and Lucas Hosseini and Hervé Jégou},
year={2024},
eprint={2401.08281},
archivePrefix={arXiv},
primaryClass={cs.LG}
}