Word embeddings for machine learning scoring on medical examination short answer items

• In Event: Credentialing

Abstract

The study investigates how word embeddings can be used in machine learning-based scoring algorithms to score text responses from short answer items on medical examinations. The performances of scoring algorithms using word embeddings in several machine learning approaches are compared. Finally, factors affecting the efficiency of these algorithms are discussed.

Authors

• Yooyoung Park, NBOME

• Xia Mao, NBOME