FAQ Search Application with FastAPI and Sentence-BERT

This repository contains a simple FAQ search functionality using FastAPI and Sentence-BERT. The application consists of a FastAPI backend that processes search queries and a frontend built with HTML, CSS, and JavaScript to interact with the backend.

Table of Contents

• Installation
• Project Structure
• Running the Application
• Explanation

Installation

1. Clone the Repository

git clone https://github.com/your-username/faq-search-app.git
cd faq-search-app

2. Set Up the Backend

Make sure you have Python 3.7+ installed.

Install the required Python packages:

pip install fastapi uvicorn sentence-transformers torch

3. Frontend Setup

The frontend is a simple HTML file. No additional dependencies are needed.

Project Structure

.
├── main.py              # FastAPI backend handling search functionality
├── index.html           # Frontend for the FAQ search
├── README.md            # Project documentation
└── requirements.txt     # Python dependencies (optional)

Running the Application

1. Start the FastAPI Backend

Run the following command to start the backend server:

uvicorn main:app --reload

The backend will be available at http://127.0.0.1:8000.

2. Open the Frontend

Open the index.html file in your browser. Make sure the FastAPI server is running, as the frontend fetches results from http://127.0.0.1:8000/search.

Explanation

This application consists of two components:

Backend (FastAPI)

• The FastAPI backend uses the Sentence-BERT model to process and find the most relevant FAQs for a given search query.
• Cosine similarity is used to match user queries to pre-embedded FAQs.
• The backend includes a single API route /search where the frontend sends search queries and receives results.

Frontend (HTML, CSS, JS)

• The frontend provides a simple user interface where users can enter a search query and receive FAQ results.
• JavaScript is used to dynamically fetch search results from the backend and display them without refreshing the page.

How It Works

1. The backend loads a predefined set of FAQ questions and their answers, embedding them using Sentence-BERT.
2. When a user submits a search query, the backend compares the query to the FAQ embeddings and returns the most relevant results using cosine similarity.
3. The frontend displays the results dynamically, with each FAQ's section, question, answer, and similarity score.