Performance Analysis of Extractive Text Summarization

Abstract

In this era, data is increasing exponentially, and it is crucial for people to keep up with all of this information. The information is available in several different forms such as news articles, online blogs, etc., many of which may be too long to read by an individual in order to gain any insight into these articles. Automatic Text Summarization (ATS) methods have been developed in order to provide important insight into these types of documents; namely, there are two types of ATS approaches: Extractive and Abstractive systems. Over time, many researchers have provided different algorithms to summarize given document raising to conduct performance analysis to determine which would be the best approach. This study has been divided into two parts: a) use of research papers that have been reviewed from IEEE and ACM libraries that were related to Automatic Text Summarization for the English language; and b) by conducting performance analysis of two well known Extractive text summarization approaches. DUC-2002 dataset has been used for validation which includes the original articles and their human written reference summaries. Both systems were evaluated by using the ROUGE evaluation matrix, ROUGE-N for bigrams.