Student Modeling for Personalized Education: A Review of the Literature - Advances in Personalized Web-Based Education - page 20

Table 1.5 Student modeling approaches in relation to student's affective features
Stereotypes Machine learning

techniques

Cognitive
theories

Bayesian
networks

Ontologies

Affective
features

6.67 % 40 % 40 % 33.33 % 6.67 %

recognition through a multi-criteria theory. Also, Conati and Mclaren (2009)
developed a probabilistic model of user affect, which recognizes a variety of user
emotions by combining information on both the causes and effects of emotional
reactions. Moreover, Moridis and Economides (2009) have developed a neural net-
work method (machine learning technique) to recognize a learner's affective state.
Also, Baker (2007) have constructed a machine learning based model that can
automatically detect when a student using an intelligent tutoring system is off-
task, i.e. engaged in behavior, which does not involve the system or a learning
task. Similarly, Cetintas et al. (2010) have performed the automatic detection of
off-task behaviors in intelligent tutoring systems using machine-learning tech-
niques. Furthermore, Balakrishnan (2011) build a student model upon ontology
of machine learning strategies in order to model the effect of affect on learning.
Machine learning techniques have been also used for predicting the emotions of
boredom and curiosity in an Intelligent Tutoring System that is called MetaTutor
(Jaques et al. 2014). Also, Hernández et al. (2010) have applied an affective
student model combining the OCC theory with Bayesian Networks. Inventado
et al. (2010) have used a combination of Bayesian Networks and machine learn-
ing techniques to model the student's affective features in POOLE III. Finally,
Crystal Island, which is a game-based learning environment, uses Bayesian
Networks to model and predict student affect for improving the learning process
and motivation.
Therefore, there are a variety of student modeling techniques that can be
used to model the learner's affective features. In Table 1.5 , the percentages of
preferences for each one of the student modeling techniques for modeling the
student's affective features are presented considering the above literature review.
The information that is derived from the particular table is the number of the
adaptive educational systems that incorporate a particular student modeling
technique for modeling the learner's emotions in a set of one hundred adaptive
educational systems. For example, if we have a hundred adaptive educational
systems 40 of them will use cognitive theories and machine learning techniques,
33.33 will use Bayesian networks etc.

1.3.5 Meta-Cognitive Features

Meta-cognitive features allow the student to be aware of her/his knowledge
and abilities and make her/him able to monitor and direct her/his own learn-
ing processes. In other words, meta-cognition concerns to the active monitoring,