Summarize4Me

Table of Contents

• Overview
• Requirements
  • Prerequisites
  • Creating virtualenv
  • Installing dependencies
• Running the program
• Next steps
• Licensing

Overview

Extractive Summarization is a Natural Language Processing (NLP) technique which aims to summarize given text based on text input. Unlike abstractive summarization, which generates new sentences based on the input, extractive summarization generates a summary from the sentences present from the original document to generate a summary.

Extractive Summarization involves four main steps:

1. Text preprocessing
2. Tokenization
3. Sentence selection
4. Summary generation

All these steps will be explained in detail in this repository.

Requirements

Prerequisites

Before running the program, the prerequisites and the virtual environment must be present. Virtual environments, in other words, are worlds of their own. Their environment does NOT affect the global environment of the system. Dependencies can be installed without interfering with the global version of the same dependencies. This allows for downgrading to be hassle-free.

Change into your desired directory and pass the following commands into your terminal:

git clone https://github.com/akshathmangudi/Summarize4Me.git
cd Summarize4Me/

Creating virtualvenv

For Python 3.6+ users:

python -m venv /path/to/virtualenv

For activation of virtualenv: bash/zsh:

$ source <venv>/bin/activate

fish:

$ source <venv/bin/activate.fish

cmd.exe:

C:\> <venv>\Scripts\activate.bat

PowerShell: 
PS C:\> <venv>\Scripts\activate.ps1

For conda users, the following commands are to be sequentially passed into your terminal:

conda -V
conda update conda
conda create -n <envname> python=x.x anaconda
conda activate <envname>

For deactivation:

conda deactivate

Installing dependencies

There's two ways to install the dependencies required for this program. Either each dependency can be installed manually (list is in requirements.txt) or a much shorter method is preferred:

pip install -r requirements.txt

Running the program

Running the program is very simple, after installation of the dependencies and creating the virtual environment, you're ready to run the application. In your terminal, simple type:

python app.py

and you'll be redirected to a localhost:8080 page with two GET blocks and one POST block.

The picture highlights a red box with the text "Try it out". This allows for the program to create the necessary files and folders to be created for prediction. This does not require any parameters.

Future action

"text" can be inputted ino the textbox and a summary will be generated based on the model training. A maximum of 2048 characters of text can be inputted and a summary upto 240 words will be generated.

Next steps

Future goals of this application is to deploy this using AWS. Thank you for checking out this project, and if you encounter any problems, please email me: akshathmangudi@gmail.com

Licensing

This repository is licensed under the MIT License. See the LICENSE file for details.