Text Summarization of a scientific document: a comparison of extractive unsupervised methods

The explosion of scientific literature and the availability of online bibliographic databases makes it extremely difficult for research communities to retrieve rapidly relevant information. Text Summarization aims to condense a given input text into a shorter version generating a reliable summary with the most relevant information about its content. Text Summarization methods can be classified into Extractive and Abstractive. This study computed three extractive unsupervised methods (TextRank, LexRank, and LSA) in order to find keysentences from a scientific document. Our goal was to compare these different approaches to evaluate which method is the most suitable for a scientific full-text.