Django-Smart-Search

Django Smart Search is a web application that allows users to easily find answers within Django documentation. It provides a link to the article and highlights the answer in the preview. The system estimates the difficulty level of every article.

---

Features

• Ask questions in natural language.
• Semantic search based on vector embeddings and BM25.
• Find answers withing Django documentation.
• Display headlines with url to relevant documentation articles.
• Results ranked according to relevance.
• Highlighted answers in the preview.
• Analysis of article's difficulty level.
• User friendly interface.

Technology Used

• Python 3.10.11
• Django (ver. 5.2.14)

Embedding Model

• Django-Smart-Search uses 'granite-embedding-english-small-r2' from Hugging Face developed by IBM team.

• The model can be found at https://huggingface.co/ibm-granite/granite-embedding-small-english-r2

Installation

Clone the repository

git clone https://github.com/Letrennia/Django-Smart-Search.git
cd django-smart-search

Create a virtual environment

python -m venv venv
venv\Scripts\activate

Install dependencies

pip install -r requirements.txt

Configuration

Create a .env file and configure the required environment variables:

SECRET_KEY_ENV=your_django_secret_key

Running the Application

Start the development server:

python manage.py runserver

The application will be available at: http://127.0.0.1:8000/

Authors

• Letrennia
• Drogon3

This project was developed as part of the Techniques and Tools of Natural Language Processing course at Bialystok Univerity of Technology.