Topic classification on short reflective writings for monitoring students’ progress

Reflection has been widely considered as an important element in student learning in higher education. Among different forms of reflective writing, one-minute papers can quickly and easily get students to reflect on their learning. Unlike short quizzes, the responses to one-minute papers could cover a wide open range and could require more time to review and summarize. When one-minute papers are administrated online, their responses are available in electronic form and this facilitates a computational approach for analysis. In this paper, we propose a machine learning approach to analyzing the students’ responses to one-minute papers. We build a text classifier to identify the topics discussed in the responses. Our results of a preliminary study conducted in a blended learning course demonstrate that the classifier can effectively detect the topics and the proposed method can be used to monitor student progress based on the detected topics. Copyright © 2017 Springer International Publishing AG.