Student Modeling for Personalized Education: A Review of the Literature - Advances in Personalized Web-Based Education - page 23

Table 1.7 Which student's characteristics are preferred for modeling
Knowledge level Error/misconceptions Cognitive features other than knowledge

52.81 % 15.73 % 40.45 %

Affective features Meta-cognitive features

16.85 % 6.74 %

According to that, the most common-modeled student's characteristic is the
knowledge level and the least common-modeled student's characteristic is her/
his meta-cognitive features (Table 1.7 ). The sum of the percentages of Table 1.7
is not 100 %. The reason for this is the fact that a system can model more than
one different student characteristics. Also, many researchers have interested in
modeling student's cognitive aspects other than knowledge. Furthermore, the
answer to the question “Which student modeling approaches are preferred in rela-
tion to student modeling characteristics?” is given in Table 1.8 . The sum of the
percentages of a line of Table 1.8 is not 100 %. The reason for this is the fact
that two different student-modeling techniques can be combined and used in
the same system. For example, a system can combine stereotypes with machine
learning techniques to model the student's learning style. The information that
is derived from each line of the particular table is the answer to the question: “if
there are one hundred adaptive educational systems how many of them will incor-
porate a particular student modeling technique to model the learner's character-
istic that corresponds to the table's line?”. The results of the research (Table 1.8 )
demonstrated that: (i) the most common used student modeling technique for the
representation of the student's knowledge level is the overlay approach; (ii) the
perturbation and constraint-based model (erroneous knowledge models) are pre-
ferred for representing the student's misconceptions and errors; (iii) uncertainty
models (like fuzzy logic techniques and Bayesian networks) and stereotypes are
preferred for modeling student's cognitive aspects other than knowledge; (iv) the
uncertainty models are, also, chosen to represent the affective and meta-cognitive
features of the student; (v) the student's emotions and affective features are very

Table 1.8 Student modeling approaches in relation to student modeling characteristics
Overlay
(%)

Stereotypes
(%)

Erroneous
knowledge
models (%)

Machine
learning
(%)

Cognitive
theories
(%)

Uncertainty
models (%)

Ontology-based
models (%)

42.55 29.79 8.51 14.89 4.26 25.53 14.89

Knowledge
level

Errors/mis-
conceptions

0 14.29 57.14 0 14.29 28.57 0

Cognitive
features
other than
knowledge

8.33 38.89 0 19.44 5.56 44.44 8.33

0 6.67 0 40 40 33.33 6.67

Affective
features