Student Modeling for Personalized Education: A Review of the Literature - Advances in Personalized Web-Based Education - page 13

Moreover, Pramitasari et al. (2009) have developed a student model ontology
based on student performance.
However, many adaptive and/or personalized tutoring systems perform mod-
eling of the student's knowledge by combining different student modeling tech-
niques. Thereby, there are systems, like TADV (Kosba et al. 2003, 2005), which
combine an overlay model with fuzzy techniques to represent the knowledge of
individual students. Stathacopoulou et al. (2005) have used fuzzy techniques to
represent the knowledge and abilities of students to help them to construct the con-
cepts of vectors in physics and mathematics. Another combination of the overlay
technique is with stereotypes. ICICLE (Michaud and McCoy 2004) is an adaptive
tutoring system that attempts to capture the user's mastery of various grammatical
units and to predict the grammar rules s/he is most likely using when producing
language by combining overlay with stereotypes. A similar combination of stu-
dent modeling techniques has been performed in ELaC (Chrysafiadi and Virvou
2013c), which is a web-based educational system that teaches the programming
language 'C'.
Also, there are adaptive educational systems that have combines overlay and
stereotypes with fuzzy techniques to model the learner's knowledge. Examples of
such systems are: INSPIRE (Grigoriadou et al. 2002; Papanikolaou et al. 2003),
which is an intelligent system for personalized instruction in a remote environ-
ment, that models knowledge on a topic classifying it to one of the four levels of
proficiency (insufficient, rather insufficient, rather sufficient, sufficient); DEPTHS
(Jeremi et al. 2012), which is an intelligent tutoring system for learning soft-

ware design patterns, models the student's mastery; and FuzKSD (Chrysafiadi and
Virvou 2014) that is an e-learning environment for the computer programming.
Other adaptive educational systems that used hybrid student models for rep-
resenting the student's knowledge level are: KERMIT (Suraweera and Mitrovic
2004), which maintains a constraint-based model and an overlay model;
InterMediActor (Kavčič

ć

2004a) that models the student's knowledge using over-
lay in combination with ontologies; F-SMILE (Virvou and Kabassi 2002) that
uses a novel combination of the cognitive theory Human Plausible Reasoning
(Collins and Michalski 1989) and a stereotype-based mechanism; and AMPLIA
(Viccari et al. 2009) that models the learner's knowledge by combining Bayesian
networks with cognitive theories. Furthermore, OPAL (Cheung et al. 2010) and
IWT (Albano 2011) used a combination of overlay and ontologies to model the
learner's knowledge level.
In addition, a variety of adaptive tutoring systems, like SimStudent (Li et al.
2011) and AIWBES (Homsi et al. 2008), used machine-learning techniques to
observe the student's behavior and make inferences about her/his knowledge
automatically. Baker et al. (2010) have used a combination of machine learning
technique with Bayesian networks in order to observe students' reactions and
adjust the instruction automatically to each individual learner. Furthermore, Web-
EasyMath (Tsiriga and Virvou 2002, 2003c), which is a Web-based Algebra Tutor,
uses a combination of stereotypes with the machine learning technique of the dis-
tance weighted k-nearest neighbor algorithm, in order to initialize the model of a

č i č