Emotion Detection based on Human PulseSignal for Supporting Teachers to Conduct Interactive Learning with Students on Online Discussion Board

• Main Authors: tai hung li, 李泰鋐
• Other Authors: Chih-Ming Chen
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/52893149808400292675

碩士 === 國立花蓮教育大學 === 學習科技研究所 === 96 === In recent years, E-learning has been a popular learning mode due to the fast growth of the Internet and it has advantages in terms of high interaction, getting feedback immediately, and breaking the limitations of learning time and space. In addition, many studies indicated that the variations of learning emotions have key affection to the learning outcomes of E-learning and many studies also proposed that detecting human emotions by physiological signals is a practicable scheme.Accordingly, the study employed sensor, signal processing, communication and system on chip (SOC) techniques to develop a embedded human emotion detection system based on human pulse signals, which can detect three human emotions including nervous, peaceful, and joyous for supporting teachers to conduct interactive learning with students on online discussion board. There are totally ten volunteers who were invited to participate in this experiment. In the experiments, several selected web movies and computer games were applied to cause emotion responses. Meanwhile, the pulse signals caused by emotion variations are retrieved by the developed embedded human emotion detection system and stored in the database for emotion analyses. To process the pulse signals for emotion detection, the extracted human pulse signals are first transformed by Fourier transform from time domain to frequency domain, then the transformed data is used to extract emotion features for training an emotion detection model by support vector machine (SVM). The accuracy rate of the modeling emotion detection mechanism evaluated by cross validation is 76.8254%. To further filter out noisy human pulse data, the accuracy rate of emotion detection evaluated by cross validation can be promoted from 76.824% to 79.7136%. Currently, the proposed human emotion detection mechanism has been successfully applied to the online discussion board to support teachers for conducting interactive III learning with students.