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

The deepMINE system uses the natural language processing based text mining from available data sources. One user enters keyword(s) related to articles he or she is interested with, system returns all the articles having the entered keyword(s) in the article title. For returning more precise articles, we have currently focused on the title of the available research articles. The system returns found research articles having the keyword(s) in the title with total count and the user can explicitly visit to abstract and summary to the research article. The deepMINE service is openly available on URL https://deepmine.in/ As discussed under Section -2, CORD-19 dataset contains nearly 1,46,115,136 words in title, abstract and article body. The below Table-2 represents the dataset category wise title word count. While mining through the deepMINE given a keyword "CORONAVIRUS" we found 2116 number of research articles as shown in Figure- It is utmost significance for a researcher to take a quick idea of research with the abstract of the research article. deepMINE system derived abstract for all mined articles from the dataset for a quick review. The below Figure-4 shows the abstarct of one of the research articles mined suing keyword "CORONAVIRUS". The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.30.014555 doi: bioRxiv preprint From the homepage of the system as shown in Figure-2 , we mined research articles using multiple keywords i.e. "CORONAVIRUS" and "RNA". This mining returns us total 98 results. Statistics of country wise research articles as shown in Figure-5 and abstract of one of the research articles is shown in Figure-6 . The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.30.014555 doi: bioRxiv preprint