Similarity-Matching-System-For-Invoices

Document Similarity Matching System

Project Overview

This project aims to develop a document similarity matching system for invoices. The system takes an input invoice in PDF format and compares it to a database of existing invoices. The goal is to identify the most similar invoice in the database based on content and structural similarity.

Approach

Document Representation

1. Text Extraction:

   • We use the PyPDF2 library to extract text content from PDF files. This involves reading the PDF file and converting its content into a textual representation.

2. Feature Extraction:

   • From the extracted text, we focus on relevant features such as keywords, invoice numbers, dates, and amounts. These features help in comparing the content of different invoices.

Similarity Calculation

1. Cosine Similarity:
   • We use cosine similarity to compare the textual content of invoices. Cosine similarity measures the cosine of the angle between two non-zero vectors in a multidimensional space, which in this case are the feature vectors of the invoices.

File Structure

• main.py: The main script to run the application.
• extract_text.py: Contains the function to extract text from PDF files.
• similarity.py: Contains the function to calculate similarity between documents.
• database.py: Sets up the in-memory database of invoices.
• README.md: This documentation file.
• invoices: Directory where existing invoices are stored.

How to Run the Code

Prerequisites

• Python 3.6 or higher
• The following Python libraries:
  • PyPDF2
  • scikit-learn
  • tfIdfInheritVectorizer

Setup Instructions

1. Clone the Repository:

   git clone https://github.com/yourusername/DeepLogicAI_Task.git
   cd DeepLogicAI_Task
   

2. Install Dependencies:

-> pip install PyPDF2 scikit-learn -> pip install tfIdfInheritVectorizer

3. Add Sample Invoices: Place your existing invoices (PDF files) in the directory. Place the sample invoice you want to compare in the invoices/directory.

Explanation of Code

-main.py:- ->Function: This is the main entry point of the application. It loads the database, extracts text from the input PDF, calculates similarity with existing invoices, and prints the results.

-extract_text.py:- ->Function: This file contains the extract_text_from_pdf() function, which uses the PyPDF2 library to extract text from PDF files.

-similarity.py:- ->Function: This file contains the calculate_similarity() function. It uses cosine similarity to compare the text content of invoices.

-database.py:- ->Function: This file sets up our in-memory database. The get_invoices() function loads existing invoices into memory for comparison.

-Further Improvements:- ->Optimization: Optimizing the similarity algorithm for better performance. ->Database Support:- Adding more robust database support for storing and retrieving invoices. ->GUI: Developing a graphical user interface for easier interaction with the system.

-Contact:- For any questions or further information, feel free to contact me at [dharmiksompura1212@gmail.com] .