An idea based on sequential pattern mining and deep learning for text summarization

Abstract

Abstract One of the Natural Language Processing (NLP) studies that has been widely researched is automatic text summarization. There are a lot of techniques and methods that are proposed for text summarization. However, not much attention has been given on the coherence and cohesion in text. The aim of this study is to present an idea to combine Sequential Pattern Mining (SPM) and Deep Learning (DL) for better text summarization process and result. In text summarization, it is important to produce understable and readable summary, and SPM as text representation extracting algorithm is capable to maintain the meaning of text by giving attention of the order of words appearance. Whereas DL is a popular and powerful machine learning technique widely used recently in various data mining studies. This study uses descriptive research methodology that collects all of the facts and information which are related to SPM and DL for text summarization, where NLP as the body of knowledge, SPM and DL as the method, and text summarization as the domain problem that need to be solved. The findings of the study are presented as a logical design and mapping of current text representation that can be implemented to further improve automatic text summarization results, in particular, to improve its coherence and cohesion.