A Novel Hybrid Student Model for Personalized Education - Advances in Personalized Web-Based Education - page 64

overlay model has been performed in ELaC (Chrysafiadi and Virvou 2013c).
Moreover, AUTO-COLLEAGUE (Tourtoglou and Virvou 2008, 2012) performs
student modeling through a hybrid student model based on perturbation and the
stereotype-based modeling technique. A combination of fuzzy logic and
machine learning techniques has been used in ADAPTAPlan (Jurado et al.
2008), while overlay model has been combined with ontologies in Personal
Reader (Dolog et al. 2004), OPAL (Cheung et al. 2010) and IWT (Albano
2011). It is remarkable to refer that a compound student model can include
more than two student modeling techniques. For example, Surjono and Maltby
(2003) have combined an overlay model with perturbation technique and stereo-
types; Chrysafiadi and Virvou (2014) have combined fuzzy techniques with ste-
reotypes and overlay model; the student model of INSPIRE (Grigoriadou et al.
2002; Papanikolaou et al. 2003) combines stereotypes and an overlay model
with fuzzy logic techniques; and the student model of DEPTHS (Jeremi et al.

2012) is a combination of stereotype and overlay modeling with fuzzy rules.
Conclusions about the most common combination of student modeling tech-
niques are drawn considering the hybrid student models of the literature review.
An overlay student model usually is combined with stereotypes or fuzzy logic
techniques. Stereotypes are blended, mainly, with overlay, but they are also
combined with machine learning or fuzzy logic techniques. Perturbation stu-
dent model is combined only with overlay and stereotypes. Machine learning
techniques are used mostly to support stereotype modeling, but there is, also,
an interest to combine them with Bayesian networks. Cognitive theories are
applied with stereotypes and Bayesian Networks. Fuzzy logic is usually used
with overlay or stereotype student models. Bayesian networks are blended,
mainly, with machine learning techniques and cognitive theories, but they are,
also, combined with stereotypes. Ontologies are primarily combined with overlay
student modeling.

ć

3.3 The F.O.S. Hybrid Student Model

A hybrid student model, which brings together various features of different
techniques of user modeling, is the solution for offering a more adaptive
learning system. The reason for this is the fact that the student model needs
to combines various aspects of student's characteristics that is both domain
dependent and domain independent in order to carry out the personalization
efficiently (Yang et al. 2010). This way, the model not only can exhibit unique
individual characteristics and preferences of each learner by monitoring
and tracing the changes of their knowledge, skills, interests, but also classify
the learners according to their performance, individual learning behaviors
and activities (Yang et al. 2010). That is the reason for the development of a
novel hybrid student, which combined overlay technique and stereotypes with
fuzzy logic.