TextSummarizer

Source code of the paper Extractive Summarization using Deep Learning

Description: Used two layers of Restricted Boltzmann Machine as a Deep Belief Network to enhance and abstract various features such as named entities, proper nouns, numeric tokens, sentence position etc. to score sentences then selecting the top scores, hence producing an extractive summary.

Prerequisites

• Python 2.7+
• Numpy
• Scipy
• Theano
• NLTK

Getting Started

1. Clone this repository
   git clone https://github.com/vagisha-nidhi/TextSummarizer.git
   
2. Place your sample text files in ./articles folder. Some sample articles are already kept for convenience.
   
3. Run python Summarizer.py to summarize the articles.
   
4. You will get the summarized outputs in ./outputs folder.

Example

Sample input and output examples have been given in this along with enhanced feature matrix for the articles.