Learning state recognition technology based on sliding filter method
Zhang Zhaohai1, Han Laiquan1, 2, Shan Mingqi1
1. Northeastern University at Qinhuangdao, Institute of Computer and Communications Engineering, Qinhuangdao, Hebei 066004; 2. Northeastern University Software School, Shenyang 110169
Abstract：In this paper, a learning state recognition technique based on sliding filter is proposed. Through the processing of feature extraction, model training and state classification of the three-axis acceleration data generated in the user's writing process, three learning indicators, i.e. the proportion of writing time, the proportion of very short writing time and the proportion of distracted state, are analyzed to identify the user's current learning state. In the experiment, the user's learning state is divided into four categories, which achieves 93.75% recognition accuracy. After applying the system to the process of classroom questioning and in-class testing, it can effectively help teachers master the learning status of each student, thus taking care of the learning progress of each student.
2019, Vol. 20 
