Detecting Learners' Hesitation in Solving Word-Reordering Problems with Machine Learning for Classification
Abstract: The aim of our present study is to develop a system to apprehend when learners hesitate in the process of solving English word-reordering problems. Learners' hesitation would be, when detected, an important clue to know learners' lack of knowledge needed to solve the problems. Our system provides an interface with which learners perform tasks of reordering given words by the use of mouse drag-and-drops, and it also records learners' study logs of mouse movements and behaviors. In this study, regarding the problem of detecting hesitation as a classification problem of study logs into two labels, or, "hesitating" and "not hesitating," we will adopt a supervised learning technique in machine learning. Parameters of mouse behavior (e.g., average speed, and drag-and-drop time) are used as features for classification, and those classified as "hesitating" behavior help teachers identify the difficulty learners have experienced in the solving process.
Presider: Kurt Ackermann, Hokusei Gakuen University Junior College