deepMINE - Natural Language Processing based Automatic Literature Mining and Research Summarization for Early-Stage Comprehension in Pandemic Situations specifically for COVID-19

Bhrugesh Joshi; Vishvajit Bakarola; Parth Shah; Ramar Krishnamurthy

In computer science, text summarization is a process of shortening the large text document(s) in order to generate short and meaningful piece of text. The objective is to create fluent natural language text keeping major insights or technicality of the source data. The automatic text summarization is an ordinary problem in the field of natural language processing and machine learning. The task was first carried out in form of generating automatic literature abstracts in 1958 [11] . Over the past half a century, the problem of text summarization has been addressed with verities of perspectives. Primarily, the task of summarization is divided into two major categories as Extractive summarization and Abstractive summarization [12] . As the name itself suggests, the Extractive method involves pulling of key phrases from the source document in order to generate the targeted summary. The Abstractive summarization works similar as we human do [13] . It involves the end-to-end deep learning technique called Sequence-to-Sequence learning to derive the understanding about the association between words. The primary objective of the our system is to deliver quick and efficient search from a huge amount of available literature. By entering a keyword CORON-AVIRUS we are getting 2116 number of research articles that contain the keyword in the title. This is even time consuming to go through the abstract of each literature, here we are interested in. Hence we found that searching for an article is not serves the ultimate purpose of serving important research articles to a user so that the researcher can speed up the research in epidemic situations. To overcome this limitation we developed a research text summarizer that can generate a technical summary by scanning all the research articles derived from user-entered keyword(s). In the demanding situation of COVID-19, we applied the literature mining with user entered keyword(s) and automatic generation of brief summary of research articles, that user searches for. The ultimate objective of our system deepMINE, is to provide quick and efficient access of the openly available research articles.