Development and evaluation of a real-time analysis method for free-description questionnaire responses

Abstract

One of the challenges of conducting collaborative and interactive classes using ICT is developing a highly functional class support system. In this study, we report a questionnaire response analysis system (QRAS) that can classify and search students’ opinions (questionnaire responses) in class in real time. The QRAS obtains distributed representations of questionnaire responses on the basis of Wikipedia Entity Vectors and implements a hierarchical/non-hierarchical clustering function and a search function, which searches for response sentences containing similar content. In addition, we add a function to determine the polarity of free-description questionnaire responses on the basis of the Japanese Sentiment Polarity Dictionary to the QRAS. The QRAS is educational support software that can be very useful for both teachers and students because it can classify and display all students’ opinions and statuses of understanding in real time from free-description questionnaire responses.