Student Modeling for Personalized Education: A Review of the Literature - Advances in Personalized Web-Based Education - page 19

1.3.4 Affective Features
The emotional state of a student affects the learning process and the student's
performance and progress. The emotional state can have a negative or positive
effect on learning. That is the reason why in real classroom settings, experienced
teachers and professors observe and react accordingly to the emotional state of the
students in order to motivate them and improve their learning process (Johnson
et al. 2000; Lehman et al. 2008). Therefore, adaptive and/or personalized educa-
tional systems should detect the emotional state of students and adapt its behavior
to their needs, giving an appropriate response for those emotions (Katsionis and
Virvou 2004).
These emotional factors that influence learning are called affective factors. The
affective states can be the following: happiness, sadness, anger, anxiety, interest,
fear, boredom, frustration, distraction, confusion, tiredness, indifference, concen-
tration and enthusiasm. Some of these emotions, like happiness and concentration,
have positive effect on the learning process. However, other emotions, like bore-
dom, tiredness and distraction, have negative effect on the learning process and
lead students to an off-task behavior (Rodrigo et al. 2007), which are associated,
usually, with deep motivational problems (Baker 2007). Off-task behavior means
that students' attention becomes lost and they engage in activities that have any-
thing to do with the learning process and aim (Cetintas et al. 2010), like surfing
the web, devoting time to off-topic readings, talking with order students without
any learning aims (Baker et al. 2004). Therefore, the affective factors should be
considered when a student model is built.
Many researchers have used cognitive, pedagogical and psychological theories
in combination with student modeling techniques in order to identify and model
the emotional states of students. In particular, Conati and Zhou (2002) have used
the OCC cognitive theory of emotions (Ortony et al. 1988) for recognizing user
emotions for their educational game prime climb. The same theory has also been used
in a Mobile Medical Tutor (MMT) for modeling possible states that a tutoring agent
may use for educational purposes (Alepis and Virvou 2011). The same researchers
have constructed user stereotypes concerning the users emotional behavior while
they interact with computers (Alepis and Virvou 2006). VIRGE is another ITS-
game, which has adopted OCC theory in order to provide important evidence
about students' emotions while they learn (Katsionis and Virvou 2004; Virvou
et al. 2005).
A significant attempt to recognize and convey emotions in order to enhance
students' learning and engagement have been done by Muñoz et al. (2010, 2011) in
PlayPhysics, which is an emotional game-based learning environment for teaching
physics. They have used Bayesian networks in combination with the Control-Value
theory (Pekrun et al. 2007), which is an integrative framework that employs
diverse factors, e.g. cognitive, motivational and psychological, to determine
the existence of achievement emotions. Furthermore, Alepis et al. (2008) have
described a novel mobile educational system that incorporates bimodal emotion