An Empirical and Comparative Study of Graph based Summarization Algorithms

Textual content is growing exponentially on the Internet, and it is also available in the form of countless archives of news articles, scientific studies, legal papers, and other materials. To summarize a text by hand takes a considerable time, a lot of effort, and a great deal of money. As a result, the use of automated text summarization (ATS) or summarization is becoming more popular. Researchers have been working on developing ATS techniques since the 1950s. Text summarization is a branch of natural language processing (NLP) study areas that uses various NLP tools and techniques to provide us with a coherent and concise explanation of large text documents. Extractive, abstractive, and hybrid approaches are used to summarize text. The TextRank, LexRank, and Modified TextRank algorithms are compared in this article. These algorithms are based on graph-based extractive approach in which sentences are represented by vertices and the cosine similarity serves as edge between two sentences. The BBC and CNN news datasets has been used for this study.