A review of research on machine learning in educational technology

Abstract

ABSTRACT The purpose of this review is to investigate the trends in the body of research on machine learning in educational technologies, published between 2007 and 2017. The criteria for article selection were as follows: (1) study on machine learning in educational/learning technologies, (2) published between 2007–2017, (3) published in a peer-reviewed outlet, and (4) an empirical study, literature review, or meta-analysis. Eighty-nine articles were chosen, after the first round of the article selection process. Through a second look at the articles, fifteen articles that did not match the criteria were eliminated. After the close examination of the seventy-four articles, certain demographical and thematic trends emerged. The top contributors to the body of research were Taiwan and the United States while the most productive year was 2017. The most utilized machine learning methods were vectors and decision trees. Commonly researched areas, on the other hand, were automation, cognitive process assessment, prediction, intelligent tutoring systems, and opportunities and challenges in the use of big data & learning analytics. Recommendations for future research focus on expanding geographical diversity, incorporating Bayesian and fuzzy logic methods more in educational machine learning work.