Semantic Search Engine using ChromaDB

---

Description

This project is a Semantic Search Engine built with Python, Sentence Transformers, and ChromaDB.

It extracts text from a PDF, creates overlapping text chunks, converts them into embeddings, stores them in ChromaDB, and allows users to search the document using natural language queries.

---

Features

• Extract text from PDF files
• Split text into overlapping chunks
• Generate embeddings using Sentence Transformers
• Store embeddings in ChromaDB
• Semantic similarity search
• Retrieve page numbers for search results
• Persistent vector database

---

Technologies Used

• Python
• PyMuPDF (fitz)
• Sentence Transformers
• ChromaDB

---

Folder Structure

Semantic_Search_Engine/

├── data/ │ └── sample.txt │ ├── chroma_db/ │ ├── index_pdf.py ├── search.py ├── requirements.txt ├── README.md └── .gitignore

---

Installation

Clone the repository:

git clone <repository-url>

Move into the project directory:

cd Semantic_Search_Engine

Install dependencies:

pip install -r requirements.txt

---

PDF Setup

This repository does not include a PDF file.

1. Place your PDF inside the data folder.
2. Rename it to:

ai_ml.pdf

Example:

data/
    ai_ml.pdf

---

Usage

Step 1: Index the PDF

python index_pdf.py

Step 2: Search the document

python search.py

---

Example

Question:

What is Deep Learning?

Output:

Page: 12

Deep Learning is a subset of Machine Learning...

---

Future Improvements

• Support multiple PDF files
• Build a Streamlit web interface
• Add hybrid search
• Integrate with Large Language Models (RAG)