13 Smart Learning Using Big and Small Data for Mobile and IOT e-Health
621
example, temporal sequence of the alarms, in order to identify the root cause of the
fault. Furthermore, the control center operators need to deal with incomplete and
inconsistent information due to data loss or additional errors. The user is modeled
by his conclusions about the faced problem, and as domain model an existing expert
system is used. This expert system [86] is in use as a decision support system at
a real power plant to identify operators’ misconceptions and to assist in problem
solving with hints by monitoring a student’s learning process. The main components
of this expert system were written in Prolog, a knowledge representation language
that had a huge impact in AI.
The tutoring module, named as curriculum planning module, utilizes case-based
reasoning to find a problem fitting the target student’s needs. In order to increase
the coverage of scenarios the ITS can handle, the knowledge base consists of both
situations from a real control center and artificial cases. A rule-based system in
the backend is implemented to adapt the level of difficulty of the content, taking
the student’s global knowledge level and the student’s learning rate into account.
The ITS also leverages an artificial neural network to calculate the distance for
case-based reasoning. Given a student’s knowledge level, it calculates the adequacy
for every case in parallel. This setup is a good example of how the two different
AI techniques can be integrated. The CoopTutor also adopts constrained-based
modeling, a mathematical approach in which decisions have to be made in a rulebound system, in order to accommodate the cooperation aspect in a multiagent
system. All important roles of the control center are either simulated or taken by
students. When educating a team of students, the interaction pattern is analyzed by
the system to ensure everybody is participating in the training [85].
In terms of knowledge representation, typical techniques for student modeling
include ontologies [87–89] and fuzzy logic [90, 91]. The same techniques are
commonly used for the domain model. A few student models have used artificial
neural networks [92]. ITS systems commonly apply the same technique for both
model types [87–91]. This is due to the increased implementation effort when
deploying multiple techniques in a single system.
An exception is provided in [93], where fuzzy logic is used for domain modeling
and a combination of fuzzy logic and rules is used for student modeling. This
approach exhibits no explicit tutor model. The tutor functionality in this ITS is
implicitly implemented by the combination of the other two models. The tutoring
modeling commonly adopts techniques such as fuzzy logic [87, 90, 91] and casebased reasoning [91, 94]. The extensive use of soft computing approaches is an
indication of the elusiveness of this problem.
13.3.3 Challenges Facing the Development of ITS
The development of an intelligent tutoring system has confronted several challenges.
First, the development process is usually a very long-term task. The development of
the above-presented system by [85], for example, spanned longer than 10 years.
621
example, temporal sequence of the alarms, in order to identify the root cause of the
fault. Furthermore, the control center operators need to deal with incomplete and
inconsistent information due to data loss or additional errors. The user is modeled
by his conclusions about the faced problem, and as domain model an existing expert
system is used. This expert system [86] is in use as a decision support system at
a real power plant to identify operators’ misconceptions and to assist in problem
solving with hints by monitoring a student’s learning process. The main components
of this expert system were written in Prolog, a knowledge representation language
that had a huge impact in AI.
The tutoring module, named as curriculum planning module, utilizes case-based
reasoning to find a problem fitting the target student’s needs. In order to increase
the coverage of scenarios the ITS can handle, the knowledge base consists of both
situations from a real control center and artificial cases. A rule-based system in
the backend is implemented to adapt the level of difficulty of the content, taking
the student’s global knowledge level and the student’s learning rate into account.
The ITS also leverages an artificial neural network to calculate the distance for
case-based reasoning. Given a student’s knowledge level, it calculates the adequacy
for every case in parallel. This setup is a good example of how the two different
AI techniques can be integrated. The CoopTutor also adopts constrained-based
modeling, a mathematical approach in which decisions have to be made in a rulebound system, in order to accommodate the cooperation aspect in a multiagent
system. All important roles of the control center are either simulated or taken by
students. When educating a team of students, the interaction pattern is analyzed by
the system to ensure everybody is participating in the training [85].
In terms of knowledge representation, typical techniques for student modeling
include ontologies [87–89] and fuzzy logic [90, 91]. The same techniques are
commonly used for the domain model. A few student models have used artificial
neural networks [92]. ITS systems commonly apply the same technique for both
model types [87–91]. This is due to the increased implementation effort when
deploying multiple techniques in a single system.
An exception is provided in [93], where fuzzy logic is used for domain modeling
and a combination of fuzzy logic and rules is used for student modeling. This
approach exhibits no explicit tutor model. The tutor functionality in this ITS is
implicitly implemented by the combination of the other two models. The tutoring
modeling commonly adopts techniques such as fuzzy logic [87, 90, 91] and casebased reasoning [91, 94]. The extensive use of soft computing approaches is an
indication of the elusiveness of this problem.
13.3.3 Challenges Facing the Development of ITS
The development of an intelligent tutoring system has confronted several challenges.
First, the development process is usually a very long-term task. The development of
the above-presented system by [85], for example, spanned longer than 10 years.