Techniques, Technologies and Patents Related to Intelligent Educational Systems - E-Infrastructures and Technologies for Lifelong Learning - page 17

Table 2. Indicative uses of AI techniques and technologies in IESs

AI Technique(s) Indicative use in IESs

Structured and relational
schemes

Representation of structural and relational knowledge
Description of learning content semantics
Definition of learning context
Exploitation for automatic creation of metadata for learning material, management of (collaborative)
learning material authoring
Description of heterogeneous, distributed Web-based learning resources.

Rule-based reasoning Representation of general domain knowledge in the form of rules

Used in most pedagogical tasks when classification tasks need to be performed

Case-based reasoning Representation of empirical (i.e. practical) knowledge

Instructional tasks (e.g. modeling of teaching strategies, learning contents adaptation according to learner
characteristics, teaching how to solve problems based on previous similar ones, constructivism).
Modeling of learner characteristics, user model management either in an environment of collaborating
IESs or in a distributed learning environment.

Neural networks Implicit representation of empirical knowledge with high levels of generalization

Classification and clustering facilities for online pedagogical tasks (e.g. learner evaluation)
Classification and clustering facilities for offline analysis of data

Bayesian networks Representation of uncertainty

Uncertainty model development in user modeling and evaluation

Fuzzy methods Representation of uncertainty, vagueness

Domain knowledge representation (e.g. relevance degrees, fuzzy metadata, imprecise links)
Domain knowledge construction (e.g. construction of concept maps)
User modeling and evaluation
Online pedagogical tasks
Analysis of data collected during IES operation (e.g. fuzzy clustering of learner data)

Constraint-based modeling Representation scheme for learner knowledge and evaluation

Genetic algorithms Evolution techniques to find adequate solutions to problems

Online pedagogical tasks (e.g. course planning, learning content selection)
Offline tasks (e.g. analysis of learner data, optimization of IES modules and contents)

Reinforcement learning The system is trained to choose sequences of actions maximizing cumulative reward

Instructional strategies tailored to learners avoiding acquisition of extensive pedagogical knowledge

Hybrid KR&R techniques Combination of more than one KR&R technique

Various functionality depending on combined techniques

Data Mining Extraction of knowledge from large volumes of stored data

Enlightenment of learning process aspects,
Performing of IES module tasks (e.g. pedagogical module tasks)
Assistance in designing/developing/refining IES modules.

Agent-based technology Autonomy

Ability to perceive information
Ability to learn from perceive information, flexibility, dynamic adaptation
Pedagogical agents
Multi-agent approaches (e.g. implementation of IES modules, modeling situations occurring in conven-
tional classroom instruction)

• Whether the approach is an IES shell/gen-
erator. Shells/generators have been em-
ployed in ITS context inspired from the
software design drive to write general/re-
usable software. Shells/generators consist
of software architectures, code libraries

and/or conceptual frameworks to make ITS
construction more efficient (Murray, 1999).
Patents (Goodkovsky, 2004; Goodkovsky,
2006) involve an IES shell/generator.
• Whether the IES involves natural language
processing or dynamic natural language