State-of-the-art approach to extractive text summarization: a comprehensive review

With the rapid growth of social media platforms, digitization of official records, and digital publication of articles, books, magazines, and newspapers, lots of data are generated every day. This data is a foundation of information and contains a vast amount of text that may be complex, ambiguous, redundant, irrelevant, and unstructured. Therefore, we require tools and methods that can help us understand and automatically summarize the vast amount of generated text. There are mainly two types of approaches to perform text summarization: abstractive and extractive. In Abstractive Text Summarization, a concise summary is generated by including the salient features of the input documents and paraphrasing documents using new sentences and phrases. While in Extractive Text Summarization, a summary is produced by selecting and combining the most significant sentences and phrases from the source documents. The researchers have given numerous techniques for both kinds of text summarization. In this work, we classify Extractive Text Summarization approaches and review them based on their characteristics, techniques, and performance. We have discussed the existing Extractive Text Summarization approaches along with their limitations. We also classify and discuss evaluation measures and provide the research challenges faced in Extractive Text Summarization.