Machine-Driven Literature Classification: A Computer-Code-Free Software to Cultivate Equitable Access to Data Science Tools

• In Event: Tools for Data Collection and Analysis in Mixed-Methods Research

Abstract

The constant expansion of published academic studies represents an important challenge for conducting systematic, reproducible, and comprehensive literature reviews. The International Association of Scientific, Technical and Medical Publishers, which accounts for the most complete data base of published research, shows that as of 2009, the number of published articles per year was 1.5 million articles. This number doubled in less than a decade, reaching 3 million academic publications in a year in 2018 alone. Although the prevalence of published literature depends on the specific topic of interest, publication growth still represents an important analytic hurdle. This study offers an analytic framework and its corresponding free software tool to identify the most salient topics characterizing published literature based on machine learning.

Author

• Manuel S. Gonzalez Canche, University of Pennsylvania