An Optimized Abstractive Text Summarization Model Using Peephole Convolutional LSTM

Rahman Rahman,Siddiqui Siddiqui

doi:10.3390/sym11101290

Rahman Rahman, Siddiqui Siddiqui

Open Access

https://doi.org/10.3390/sym11101290

Copy DOI

Journal: Symmetry	Publication Date: Oct 14, 2019
Citations: 18	License type: CC BY 4.0

Affiliation: Dhaka University of Engineering & Technology

Abstract

Abstractive text summarization that generates a summary by paraphrasing a long text remains an open significant problem for natural language processing. In this paper, we present an abstractive text summarization model, multi-layered attentional peephole convolutional LSTM (long short-term memory) (MAPCoL) that automatically generates a summary from a long text. We optimize parameters of MAPCoL using central composite design (CCD) in combination with the response surface methodology (RSM), which gives the highest accuracy in terms of summary generation. We record the accuracy of our model (MAPCoL) on a CNN/DailyMail dataset. We perform a comparative analysis of the accuracy of MAPCoL with that of the state-of-the-art models in different experimental settings. The MAPCoL also outperforms the traditional LSTM-based models in respect of semantic coherence in the output summary.

Highlights

At present, it is a challenging task to retrieve useful information from a large text due to the unseen growth of blogs, article, and reports
The goal of this study is to develop an abstractive text summarization model using a variation of long short-term memory (LSTM) called peephole convolutional LSTM
The parameters of the developed model are optimized to improve the model performance using the central composite design (CCD) in combination with the response surface methodology (RSM)

Summary

Introduction

It is a challenging task to retrieve useful information from a large text due to the unseen growth of blogs, article, and reports. An automated text summarization technique gives an effective solution to extract useful information from the large text. Text summarization is a task of condensing a text that keeps the actual meaning and important parts of the original text. A concise and quality summary assists humans in processing and understanding a large text in a short time. Automatically summarizing a large text is still an open problem. There are two possible ways to perform text summarization: extractive and abstractive [1]. The process of copying words and sentences directly as a summary from the long text is called extractive summarization. Most of the conventional text summarization models are based on the extractive text summarization (ETS)

Objectives

Methods

Conclusion