Techniques, Technologies and Patents Related to Intelligent Educational Systems - E-Infrastructures and Technologies for Lifelong Learning - page 16

al., 2003) an AEHS uses neural and fuzzy modules
in domain knowledge, learner evaluation, and
pedagogical module. In (Koutsojannis et al., 2007)
a combination of RBR with a genetic algorithm
approach to determine the difficulty levels of the
provided exercises is described. In (Bittencourt,
Tadeu & Costa, 2006) an agent-based IES com-
bining RBR and CBR is presented. In (Huang,
Huang & Chen, 2007) an approach combining a
genetic algorithm with CBR is presented.
Application of data mining to IESs is a recent
trend. Useful knowledge may be extracted from
large volumes of data stored in IESs (especially
Web-based ones). Data mining is a multidisci-
plinary area exploiting methods from fields such
as AI, Machine Learning, Statistics and Data
Bases. There are several data mining techniques
that can be employed in IESs such as statistics and
visualization, prediction, classification, clustering
and outlier detection, association rule mining and
text mining (Romero & Ventura, 2007; Castro et
al., 2007). Such techniques can be addressed to
learners, tutors, authors and academics respon-
sible. They can be used in different perspectives:
to enlighten learning process aspects, to perform a
specific task of an IES module (e.g. pedagogical
module tasks) and to assist in designing/develop-
ing/refining IES modules. In the first perspective
for instance, association rules could be used to
discover associations between learners and learn-
ing material. In the second perspective, improved
learning experiences could be provided (e.g. using
clustering to group similar learners and promote
CL). In the third perspective, tools for evaluat-
ing learning content could be provided (Castro
et al., 2007).
Agent-based technology is sometimes used in
IESs. Intelligent agents are AI programs with the
ability to perceive and act upon their environment
(Russell & Norvig, 2009). Characteristics of in-
telligent agents involve autonomy and ability to
learn from perceived information. Various types
of intelligent agents can be defined reflecting the
kind of information made explicit and used in deci-

sion process. Agents can offer great flexibility and
make dynamic adaptation feasible. Furthermore,
pedagogical agents (i.e. human-like artificial
characters) would play an important role in this
direction (Chou, Chan & Lin, 2003; Alepis, Virvou
& Kabassi, 2007). Multi-agent environments can
be also defined. In such cases, the multiple agents
may either compete or cooperate. Multi-agent
approaches in IESs can prove useful in different
ways. IES components can be implemented by
employing a multi-agent approach. Employing
multi-agent approaches in IESs, we can model situ-
ations that occur often in conventional classroom
instruction (e.g. CL). For instance, agents may
implement artificial learner companions helping
human learners learn collaboratively if they want
so, even when no other peer human learners are
around (Devedzic, 2005). Another such case may
involve learning from multiple tutors (Avouris &
Solomos, 2001).
In Table 2, a summary of possible uses of AI
techniques/technologies in IESs is provided.

Solutions from Recent Patents on
IESs

A number of patents related to IESs have been
approved. Patents on IESs can be categorized
according to various views such as the following:

• Whether the IES is an ITS or an AEHS.
Most of patents on IESs involve ITSs. The
patent in (Chakraborty, 2006) involves an
AEHS approach.
• The employed AI techniques/technologies.
Such information is shown in Table 3. It
can be seen that most patents involve RBR.
• The domain type to which the IES can be
applied. Table 4 presents types of domains
to which various patents are applicable.
• Whether the IES provides authoring tools/
facilities. Table 5 presents authoring facili-
ties provided by patents.