*2.2. Affective Tutoring System*

Emotions play a significant role in human behaviors in individual and social communities. This can happen in any kind of human activity such as learning online [13]. Recently, researchers have acknowledged the role of emotions in online learning in improving learning outcomes and enhancing students' experience [14–16]. The significance of incorporating emotional states with the learning process has necessitated the development of ATSs, which is the extended research of ITSs and with the ability to adapt to the learner's adverse emotion effectively to spark the learner's motivation to learn [17]. The affective tutoring system is based on the intelligent tutoring system, combining with affective computing and featuring the ability to detect the learners' emotions when they are learning [18]. The affective computer-based digital learning system proposed by Duo and Song (2012) aims to simulate the traditional teaching mode to analyze and recognize learners' emotions and improve learners' moods with virtual agents [19]. Mao and Li (2010) proposed that success in teaching lay in the ability to quickly identify the learners' emotional state, timely adjust the learners' emotions and enhance the learners' learning motivation. Ammar et al. (2010) added a facial expression detection module to the affective tutoring system to boost the learners' moods and improve the emotional communication between the system and the learner. The final results indicate that affective computing can effectively monitor the learners' emotions, appropriately lead to positive emotions and thus improve the learning motivation [20]. As Gerald (2004) said, learners may greatly reduce their learning motivation due to negative emotions, but positive emotions can effectively improve the learners' learning willingness. Graesser et al. (2004) used natural language to set up an emotion module for about 1000 subjects who are students in computer or physics majors. As the test results showed, such an emotion module has significantly improved the learning effects on both basic knowledge learning and in-depth research and discussion [21]. The indicators for measuring the satisfaction of using the affective tutoring system include learners' attitude and affective computing, the performance ability of the tutoring system, the accuracy of emotion recognition, the quantity of emotion recognition, the teaching course activities and the availability of the system etc. [17]. This study will therefore take the above characteristics into account in the design of the system.
