Knowledge-Based Systems 33 (2012) 180–194

Contents lists available at SciVerse ScienceDirect

Knowledge-Based Systems
journal homepage: www.elsevier.com/locate/knosys

Representing instructional design methods using ontologies and rules

Christian Vidal-Castro a,⇑, Miguel-Ángel Sicilia b, Manuel Prieto c
a
Universidad del Bio-Bio, Avda. Collao 1202, 4081112 Concepción, Chile
b
Universidad de Alcalá, Ctra. Barcelona km. 33.6, 28871 Madrid, Spain
c
Universidad de Castilla-La Mancha, Po. de la Universidad 4, 13071 Ciudad Real, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Instructional design theories are design theories that offer explicit guidance on how to help people to
Received 29 September 2010 learn in specific situation. They can be used to guide the design of learning activities and the arrangement
Received in revised form 2 March 2012 of associated resources. These theories are currently expressed in natural language, but they are often
Accepted 1 April 2012
given some structure in terms of methods and conditions. Tools supporting theory-based instructional
Available online 9 April 2012
design require formal models of these theories to be expressed in languages with computational seman-
tics, thus allowing their processing. Recent research has resulted in ontologies describing theory-neutral
Keywords:
learning activity sequences and resources in accordance with proposed standards like IMS LD, but these
Ontology
Instructional design
are not sufficient for building instructional design aid tools. This paper describes the use of formal ontol-
IMS LD ogies to partially represent instructional design methods in a form that can be used to build such support-
OWL ing tools. Combining ontologies describing learning activities in a learning design with rules and
SWRL constraints, it is possible to encode some forms of instructional design. This paper describes such an
approach using OWL and SWRL. The approach has been evaluated by building a catalogue of instructional
design methods expressed in these languages, obtained from a systematic extraction from a catalogue of
instructional design theories. The practical utility of the ontological schema is demonstrated by means of
a relevant case study.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction
Diverse kinds of technologies have been used in e-learning,
some aimed at supporting learners and others at supporting teach-
ers and instructors. Regarding the former, support for students is
reflected in efforts to improve the platforms and media used to
achieve student learning by adapting them to their characteristics
and learning needs [1]. Examples include intelligent and adaptive
learning systems or collaborative learning systems. The latter use
technology to give assistance to instructors in devising learning
activities or resources, or in supporting the tutoring and assess-
ment of learning experiences. In e-learning environments, teachers
perform various activities [37] including the development and
publishing of resources in repositories or Learning Management
Systems platforms, thus helping students to solve problems and
themselves to manage the learning process and assess student per-
formance. To support these activities, several computer-supported
techniques such as knowledge models [6], design tools [4], data
mining [38] and intelligent systems [18] have been reported
elsewhere.
The design of learning resources is an essential activity in
instructional design. From the perspective of learning contents,
⇑ Corresponding author. Tel.: +56 412731273; fax: +56 412731290.
E-mail address: cvidal@ubiobio.cl (C. Vidal-Castro).
these are presented using various media such as hypertext and
multimedia (text, images, audio and video), usually structured
with a view to producing reusable learning objects (LO) [11].
According to the IEEE LOM standard an LO is any digital or non-dig-
ital entity that can be used to learn, educate or teach. The ultimate
goal of learning resources is to enable and enhance learning, but a
large amount of these resources lack any defined instructional
strategy, which causes many of them to fail in their goals [47].
The concept of Learning Design (LD) is defined as the applica-
tion of instructional design knowledge to develop a unit of learning
[25]. In another sense, ‘‘learning design’’ refers to the device result-
ing from the design process, i.e. the plan, learning resources, the
ordering of activities and the tools required to carry out a particu-
lar learning experience [24]. The IMS LD specification permits the
modelling of such learning process specifications. Through an
IMS LD instance, resources, learning objectives, prerequisites, com-
ponents such as roles, activities, environments consisting of LOs
and services, methods, plays, acts and role-parts are specified
and arranged. However, this specification does not allow the
recording in digital form of the decisions or the instructional de-
sign methods used in the construction of an LD [41].
The construction of learning resources, more particularly of LDs,
requires disciplined approaches. Instructional Engineering [32] is
one method that uses a systematic approach to producing learning
systems. Founded in Software Engineering, Knowledge Engineering

0950-7051/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.knosys.2012.04.005
 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
and instructional design, it facilitates the systematic development
of the resources by providing methods and tools which come from
software development. It offers support for the management of
development, i.e. it considers the method, its phases, its stages,
its activities, its metrics and resources involved.
All these techniques can be used to support the design process.
Instructional Engineering also uses methods from the field of edu-
cation. An example of these are the Instructional Design Theories
(IDT) methods which attempt to guide the designer in the con-
struction of learning resources. According to Reigeluth [36], an
IDT is a theory that offers explicit guidance on how to help people
to learn. Some of these theories have been used to support resource
development in e-learning [29]. Apparently, there is an ever-grow-
ing need in e-learning to use IDT to construct learning resources.
However, ID theories are currently expressed in natural lan-
guage (in books, articles, etc.) and not in the sort of computational
semantics that would make them usable when building intelligent
computer tools. If representations were available written in com-
putational semantics, they could be read by software.
Knowledge Engineering provides techniques of representation
and knowledge acquisition that are useful for design purposes.
From a broader perspective, it provides many artificial intelligence
techniques such as model-based personalisation, intelligent sys-
tems and techniques based on ontologies and semantic web [39]
More specifically, semantic web technologies can be used to build
shareable knowledge models based on open standards such as the
W3C OWL recommendation. According to Berners-Lee et al. [5],
‘‘the Semantic Web is an extension of the current web in which
information has a well-defined meaning, more understandable by
computers, and where people work cooperatively’’. In this sense,
which encompasses technologies used in recent times, ontologies
enable the meaning of things on the web to be defined better. They
are designed so that this meaning can be processed by machines
and humans on account of to their formal semantics [19]. The
use of ontologies in the context of semantic web technologies is
more oriented to delivering meaning to machines so that auto-
matic systems can read and use that knowledge. Semantics are ex-
pressed through description logics. In this context, languages based
on RDF have evolved in the last years, allowing for the sharing of
ontologies. OWL, the ontology language recommended by the
World Wide Web Consortium, exploits many of the capabilities
of description logic, including a well-defined semantics and some
techniques for practical reasoning. Another important language is
SWRL (Semantic Web Rules Language) which enables rules to be
defined for ontologies. It is a language which allows the description
of the knowledge deductible from the ontological structure by rea-
soning about rules [7]. OWL allows to build hierarchies of concepts
defined through a language of axioms, facilitates interpretation
based on reasoning, describes concepts and relationships with
other concepts. SWRL adds an additional layer of expressiveness
which enables to define rules of inference to be defined in these
models [30].
What motivates our research is the need to represent instruc-
tional theories in an interoperable and machine-understandable
form for the practical purpose of building advanced instructional
design tools. This paper presents an approach to the representation
of methods of instructional design theories using semantic web
technology. Using an ontology of the IMS LD standard as a base
representation of the design outcomes, we specify representations
in a computer-understandable form that can be used by theory-
aware software. The approach followed consists in analysing each
of the methods described as components of existing instructional
design theories and expressing them in a combination of OWL
and SWRL as far as is possible. In continuation of previous research
[44], this work provides a complete catalogue following the gen-
eral modelling directions already established. It also provides a
181
first complete case study demonstrating the applicability of the ap-
proach for practical purposes in a domain in which mature domain
ontologies are already in place.
The rest of this paper is structured as follows. Section 2 reviews
such key aspects of the background to this research as instructional
design, Educational Modelling Languages and ontological repre-
sentations of IMS-LD specification. Section 3 explains the reasons
for using ontological language to represent IDT methods. Section
4 presents the objectives, assumptions and approach used when
modelling and also describes the definition of a rule-based cata-
logue of IDT methods. Section 5 then shows fragments of represen-
tations that are part of that catalogue. Section 6 presents a case
study to demonstrate the usefulness of the catalogue. Finally, con-
clusions are drawn and scope for future work is suggested.
2. Background
This section describes aspects of instructional design, educa-
tional languages and IMS LD specification that are of importance
to understanding the approach described in the paper. In addition,
studies reporting ontological representations for IMS-LD are
surveyed.
2.1. Instructional design
Instructional Design (ID) is the systematic application of theo-
ries and principles that guide the design of learning resources. IDTs
help us to know how the body of knowledge is organised for use in
the learning process [12]. The use of IDT offers guidance for the
construction of learner resources by considering elements of the
instructional context and the learner’s learning goals. In contrast
to learning theories, ID theories are prescriptive: they are oriented
to design and identify methods and contexts of use [36]. Their
methods can be broken down into sub-methods. ID methods are
more probabilistic than deterministic, meaning that the use of
none of them ensures the achievement of objectives, but rather in-
creases the chances of achieving them. The situation of use refers
to certain elements of context that influence the choice of methods,
elements like instructional objectives and instructional conditions.
According to the goals of this research, the characteristic of divi-
sion into submethods becomes beneficial for the analysis and mod-
elling of ID methods.
Reigeluth compiled numerous instructional design theories in
‘‘Instructional-Design Theories and Models, Volume II: A new para-
digm of Instructional Theory’’, converting his work into a major
source of knowledge about these theories, which include: Learning
By Doing, Collaborative Problem Solving, Landamatics for Teaching
General Methods of Thinking, Multiple Intelligences, Instructional
Transaction theory and Elaboration theory.
Moreover, applying IDT methods requires a disciplined ap-
proach to indicating, for example, the sequence of activities and
the results of each stage. Instructional design process models allow
instructional systems to be created from a system perspective [28],
covering stages ranging from analysis to implementation and eval-
uation. Some models propose a linear sequence for these activities,
while other recommend models that consider iterations and incre-
mental developments.
One of the most widely used ID model is ADDIE (Analysis, De-
sign, Development, Implementation and Evaluation). ADDIE is re-
garded as a simplified model for learning resource construction
[34]. Other examples of these methods are ASSURE [17], ARCS
[20] and Dick et al. [9]. However, these models focus on phases
and do not provide guidance to the designer when making deci-
sions about whether a method is applicable in some context or give
details of how to perform the activities.
182 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
2.2. Educational languages and IMS-LD
Rawlings et al. [35] define an Educational Modelling Language
as ‘‘A semantic information model and binding, describing the con-
tent and process within a unit of learning from a pedagogical per-
spective in order to support reuse and interoperability’’.
Most of these languages use XML as a language binding,
although some have attribute/element owners which do not have
a high level of interoperability and reusability. There are two dis-
tinct language groups: those which do not consider the existence
of a pedagogical model, as is the case of LMML, CDF, Targeteam
and TML and those which satisfy the definition of an educational
modelling language. This latter group includes PALO and OUNL-
EML. Other languages have emerged to date, some with graphical
notations, such as MOT + [33], which is considered a visual lan-
guage that assists in instructional scenario modelling by means
of the MISA method.
OUNL-EML, or simply EML (Educational Modelling Language)
was developed by the Open University of the Netherlands with
the intention of providing a semantic notation for units of learning
for use in e-learning [23]. EML was one of the first educational
modelling languages to allowed instructional design to be inte-
grated and to express pedagogical strategies such as problem-
based learning, portfolios or collaborative learning [22].
EML introduces the concept of UoL (Unit of Learning) as the
smallest unit that provides events to meet instructional objectives.
A UoL includes learning activities, communication facilities, moni-
toring and mentoring, search and workflow [23].
OUNL-EML, and its subsequent integration into IMS-LD, is re-
garded as one of the most important initiatives in the direction
of the integration of instructional design in e-learning.
IMS Learning Design (IMS-LD) is an open standard that is used
to encode a wide range of digital courses in a formal, semantically
interoperable and machine-understandable way [24]. This specifi-
cation incorporates other specifications such as IMS Content Pack-
aging, IMS QTI and IMS RDCE, and has become the de facto
standard for describing designs and UoL [3].
IMS LD enables learning processes to be modelled and interac-
tive communication between the actors involved. It basically de-
fines who, when, where, how and what resources, services and
learning activities are to be used. It allows learning scenario to
be modelled and various pedagogical approaches to be imple-
mented. All the elements contemplated in the IMS LD specification
can be modelled using design level A, considered to be the simplest
level. The next level, B, considers the model of conditions, proper-
ties, services and global elements. Level C allows notifications to be
used.
IMS LD is agnostic to instructional design theories, so it can sup-
port the use of a wide range of e-learning pedagogies and facili-
tates the reuse of teaching practices rather than the reuse of
content [15]. It permits the result of the design process to be ex-
pressed, but not the rules, guidelines or instructional methods fol-
lowed when creating a LD [41].
As far as tools are concerned, the goal of the design process is
the creation of an LD. In general the tools are classified into author-
ing tools and execution servers. Authoring tools, or editors, provide
support for designers in the tasks of developing a LD. These tools
differ in levels of support for the specification, the use of tree-based
interfaces or graphics, and the level of compliance with the stan-
dard in the UoL generated, among others. However, all of them
generate a encoded LD in XML schema. A second group of tools
are called execution machines. These work as servers on which
the UoL is executed by users. Users assume some of the roles de-
fined in LD interacting through the implementation of activities.
Also worth mentioning is a third group of tools for managing
UoL stored in a repository. These tools are intended to provide ser-
vices to organisations or communities which want to share and re-
use the results of their LD [14]. In general, there is consensus that
the use of editors and tools for implementation of LD is not easy.
While it is true that most users need not deal with the underlying
XML structure or interfaces of the editors, their use is not yet
widespread.
2.3. IMS-LD ontologies
The IMS-LD specification was formally modelled using XML-
Schema. However, this language has difficulties of expression
when it comes to describing the semantics associated with the ele-
ments of IMS-LD [2]. These difficulties include: the explicit defini-
tion of relationship hierarchies; the formal definition of constraints
between concepts, relationships and attributes; and the establish-
ment of mathematical properties of relations such as symmetry
and transitivity and taxonomic properties such as disjoint and
exhaustive classes. The use of formal ontological language may im-
prove the expressiveness of the semantics contained in IMS-LD
specification.
The IMS-LD ontology proposed by Lama et al. [26] provides an
ontological representation of this specification. It describes the
structure and sequence of learning activities, roles, services, re-
sources, among others. The ontology includes taxonomic relation-
ships and formal axioms. The taxonomy of concepts includes
concepts such as learning design, learning objective, prerequisite,
method, play, role, role-part, act, environment, learning object
and activities. Additionally, the semantics expressed in the specifi-
cation is modelled by formal axioms expressed in first-order logic.
Another ontology that relates LDs with LOs has been presented
by Knight et al. [21] in a study adopting three ontologies: an ontol-
ogy based on IMS-LD, a LOs content ontology, and an ontology that
links up with the first two, the aim being to link LOs descriptors
with their possible contexts of use.
3. Reasons for using ontology languages to support
instructional design
The ID process for the construction of LDs is not a trivial pro-
cess. This is due, among other factors, to the nature of instructional
design and human learning. The following are some of the prob-
lems related to the learning design:
(1) Different positions or educational foundations can be used
depending on the epistemological principles or theories
used to understand human learning [43].
(2) Learning design may be regarded as a problem of ‘‘expand-
able rationality’’ [45], causing expansions of the problem in
relation to the initial requirements.
(3) Learning is not deterministic, nonstatic and nondiscrete. It is
a process that evolves, each learner building its own learning
differently. Learning is not observable, since it is a process
that occurs ‘‘in the mind’’ of each learner [45].
(4) Learning design may include numerous combinations of
materials, activities and an endless amount of usable
resources.
All this means that learning design requires open rationality
[42]. As a result, several design solutions may exist for the same
learning objective.
However, some opportunities have Arisien for incorporating
technology to support this process thanks to such factors as:
(1) The existence of a standard for expressing learning designs, a
case in point being IMS-LD. The use of IMS-LD provides:
 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
 A common language to express learning activities, which
takes account of activities, participants, roles, and
resources grouped according to a target. All components
are structured in a scheme that can be interpreted
correctly.
 High expression and correct interpretation by systems
that can run. That is to say, it can represent any learning
activity by generating XML-based schema which can be
understood by computers.
 Flexibility to support different instructional methods and
approaches such as active learning, collaborative learn-
ing, competency-based learning [24]. As an educational
language IMS-LD allows the creation of learning scenar-
ios but does not record the use of a given method in
developing instructional resources
 The LD resulting from the design process is a resource
that can be shared and compared to confront other
designs. It collects the knowledge used by the designer
to achieve a goal in a particular instruction context.
(2) It is possible to use artificial intelligence techniques to sup-
port the process in order to provide some degree of automa-
tion. More particularly, through knowledge representation it
can model the knowledge used by an expert designer and
treat this representation as a knowledge base for intelligent
systems that support the design process.
(3) The availability of IDT, which are practice-oriented theories,
can guide design with the aid of methods and guidelines
used in certain contexts. If these methods were represented
in a computable semantics they could be used by intelligent
systems to assist in design [44].
Our research points to the lack of any formal language to ex-
press ID methods used in the construction of LDs. The representa-
tion of knowledge from IDT can be useful for designers of e-
learning environments, and some studies have used ontologies as
models of this type of knowledge [42,16]. Efforts to dispose of
instructional model representations using formal ontology lan-
guage are crucial to supporting the design process through the-
ory-aware systems.
Currently there are not languages which enable ID methods to
be represented in a computationally processable language. One
ontology, called Omnibus which sought to cover a broad spectrum
of educational theories, was presented by Bourdeau et al. [8] and is
described in detail by Hayashi et al. [16]. It proposes an ontological
infrastructure that allows designers to use knowledge of learning
and instructional theories.
The ontology links a unit of learning with learning paradigms,
learning theories, and instructional theories. The ontology was
built using the Hozo editor1 and was not originally designed to
use with semantic web languages. Its classes represent multiple ele-
ments of a learning process as learning events, phases of DI models
and instructional foundations. But it is a proposal which is not fo-
cused on modelling DI methods and furthermore poses problems
of compatibility with IMS-LD standard, thus hampering interopera-
bility and use by other software.
Sicilia et al. [44], provided the starting point for developing a
language for instructional models. Their proposal used a formal
ontology to represent the outputs of design process, i.e. a LD. ID
methods were used as constraints to the structure and contents
of the ontology.
The main objective of the present work is to represent multiple
ID theories using formal ontologies, forming what we call a cata-
1
http://www.hozo.jp/, consulted on July 22th 2010.
183
logue of ID theories. The catalogue could be used as a knowledge
base for systems that require the use of knowledge regarding the
application of certain IDT. For example, when applying Reigeluth’s
sequencing theory, a theory-aware tool could guide the LD de-
signer towards using broader concepts first.
4. Assumptions and modelling approach
This section presents the goals that guide this research as well
as the assumptions used in modelling. It also sets out a formal def-
inition of the catalogue and the approach used to incorporating
representations of theories.
4.1. Goals and assumptions used in modelling
The objective of modelling is to have formal representations of
ID methods to be used by tools to guide the application of a specific
IDT. This could be achieved by building design templates or wiz-
ards, for example. Additionally, the representations could serve
as a knowledge base for systems that perform validation of confor-
mance to a particular IDT. This would determine whether a LD
meets the guidelines of the methods. Other uses of the representa-
tions are related to assistance in LD searches in repositories by
facilitating reuse of designs.
As already mentioned, on the basis of Sicilia et al. [44]’s pro-
posal, it is possible to encode instructional design methods using
a formal ontology that represents LDs. For example, the Elabora-
tion theory method ‘‘Teach broader, more inclusive concepts before
the narrower, more detailed concepts that elaborate upon them’’ can
be represented by analysing the concepts considered in the LD,
which include other concepts. However, not all guides described
in the methods can be encoded using a formal language. For exam-
ple, the method from multiple intelligence theory ‘‘Pick representa-
tions that capture important aspects of the topic’’ relies on one’s
knowing what kind of representations are better at capturing
important aspects of the topic, and this is difficult as it depends
on the interpretation of representations and topics. In conse-
quence, when we use the ontology and rule representation of the
theory for checking a concrete design (for example, one described
using IMS LD), the outcome of the checking cannot amount to full
conformance with the instructional theory, but only partial or pro-
visional conformance. We will therefore use the concept of provi-
sional conformance to refer to an LD’s compliance with the
guidelines of ID methods: ‘‘A concrete Learning Design LD, expressed
in a digital educational description language, is provisionally confor-
mant to the IDModel A if there exists a legal interpretation Li of A in
terms of the description language and LD fulfills all the constraints
contained in Li’’. This definition of conformity for a LD is tentative,
because representation methods could be improved in the future.
However, future improvements in formal languages could
strengthen the expression of legal interpretations, which might
lead to the partial or full representation in the future of a method
now classified as unrepresented.
Moreover, according to Reigeluth, IDT are composed of methods
that can be decomposed into other more specific methods. The
methods are designed to be used under certain conditions or con-
texts, but this point is not considered in this work.
For the creation of the ID methods catalogue IMS-LD ontology
will be used [2], thus enabling a LD to be represented (instantiated)
a LD. This ontology, expressed in OWL, contains descriptions of
learning activities and their structure could be used as the basis
of constraints and rules for representing IDT [44].
The IMS-LD Ontology is composed of 3 sub-ontologies:
 ld.owl: Represents concepts and relationships associated with
the organisation of the elements of an LD according to the stan-
dard IMS-LD.
184 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194

 uol.owl: Represents and organises concepts and relationships The SWRL language is used to define rules in ontologies [30].
connected with the resources associated with LD. The rules represent guidelines indicated in ID methods. Thus, ID
 lom.owl: Represents the standard IEEE-LOM metadata for learn- methods represented in the form of rules become restrictions to
ing objects used by the LD. the ontological structure of LD.

Table 1 describes some of the information stored in each sub- 4.2. Modelling approach used
ontology as well as the namespace used for each.
The review and study of source documentation (books and
other bibliography) enabled a rule-based catalogue of ID theory
to be compiled. Through a process of elicitation, based on docu-
Table 1
ment review techniques, rules were extracted that represent part
Extract the information stored in each sub-ontology.
of the methods of ID theory. The criterion for the selection of the-
LD elements Sub- Namespace ories included in the catalogue was related to the possibilities of
ontology
representation in formal language and their application in e-learn-
Acts of each play defined in LD method, which ld.owl ld ing environments. As for the methodology adopted for modelling a
contains learning activities performed by roles
theory, the activities performed are listed in Table 2.
(participants), their environments and services
Resources associated with the activities, taking uol.owl uol
account of learning objectives, learning objects, 4.3. Catalogue definition
prerequisite and feedback
Metadata information of IEEE-LOM associated with lom.owl lom The catalogue of ID theory (C) is defined as the set formed by
learning-object
the union of theories represented:
[ [
C ¼ fT 1 T2 T3    [ T ng ð1Þ
Table 2
In abbreviated form it is expressed as follows:
Activities for modelling each ID theory.
[
Activities Description C¼ Ti ð2Þ
i¼1::n
Preliminary Documentation review of the theory to be represented,
review analysing the scope for modelling for its methods. Those where Ti is the representation of each ID theory and n is the total of
methods were selected whose representation was most represented theories.
feasible
A represented theory (T) is defined as the union of all methods
Analysis Identification of those methods that could be
represented completely or partially, as well as of those that compose:
methods that are common to other theories. Discussion [ [ [
of their reuse
T ¼ fM 1 M2 M3    Mp g ð3Þ
Representation of Identification of guides described in methods are
methods identified and adoption as SWRL rules Expressed in short form:
Validation Preliminary validation of method representing rules [
Addition to the Inclusion of set of rules into catalogue
T¼ Mi ð4Þ
catalogue i¼1::p

Fig. 1. Relationship between IDT catalogue and other ontological structures.

 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
where Mi is the representation of each method of an ID theory and p
is the total of represented methods.
Method representation (M) is defined as the set formed by the
union of rules that represent their guides:
[ [ [
M ¼ fr 1 r2 r3    rqg
Expressed in reduced form:
[
M¼ ri
i¼1::q
where ri represents each rule of a method and q is the total of rules
that model the method whether completely or partially.
A feature of the catalogue is that there are rules that are com-
mon to several theories or used to represent different methods in
a same theory. This set is called common rules.
Fig. 1 shows the relationship between IDT catalogue and other
ontological structures.
5. Some representations of ID theories
This section shows fragments of IDT representation. Currently,
the catalogue includes 3 represented theories: Elaboration theory
[36], Multiple Intelligences [13] and Learning by Doing [40]. These
representations are part of the rule-based catalogue of IDT methods.
5.1. Characteristics of the rules used in the representation
The rules were constructed after considering the ontological
structure for representing the standard IMS-LD. Other ontological
structures (see Fig. 1]), which are not part of the IMS-LD ontology,
were also necessary for the rules to be represented:
 Some IDTs require a domain ontology. For example, Reigeluth’s
sequencing theory requires the identification of broader topics,
thus necessitating a domain ontology. Gene Ontology [10] is
another example of such ontologies, this time from the domain of
genetics. These ontological structures are external to the catalogue
and can be reused to support the representation of certain IDTs.
 Topics general schema. This is a schema that represents the top-
ics or concepts considered in a LD. It is related to the learning
objective that points to a topic of domain ontology. In the onto-
logical structure of the catalogue this schema is identified by
tgs (topics general schema) namespace.
 A scheme to represent the results of reasoning. For example,
greaterDepth property records a ‘‘true’’ or ‘‘false’’ indicating
whether the rule was satisfied or not. This structure is identified
by rr (rules results) namespace.
Classes, properties and restrictions defined in ontology are used.
For example, the depth of the method ‘‘Select fewer topics to treat
them in greater depth’’ from Multiple Intelligences theory was mod-
elled in relation to the terms it includes. The depth generates a
hierarchy of knowledge items related to learning objectives. For
this case a topic was considered to be of ‘‘greater depth’’ if it has
more than 2 deep levels. The following rule represents this situa-
tion. Of course, this implementation can have other interpretations
based on other considerations than concept hierarchy depth.
tgs : KnowledgeItemð?c1Þ ^ tgs : KnowledgeItemð?c2Þ ^ tgs
: KnowledgeItemð?c3Þ ^ tgs : KnowledgeItemð?c4Þ ^ tgs
: concept  includesð?c1;?c2Þ ^ tgs : concept
 includesð?c2;?c3Þ ^ tgs : concept  includesð?c3; ?c4Þ^
differentFromð?c1; ?c2Þ ^ differentFromð?c1;?c3Þ^
differentFromð?c1; ?c4Þ ^ differentFromð?c2;?c3Þ^
differentFromð?c2; ?c4Þ ^ differentFromð?c3;?c4Þ
! rr : greaterDepthð?c1;“true”Þ
185
This rule uses the concept-include property to verify that a
KnowledgeItem includes another one. The DifferentFrom ( )
predicate is included to ensure that instances are not same, accord-
ing to the open world assumption. If the rule complies, rr:great-
ð5Þ erDepth property is set to ‘‘true’’ value.
An important aspect mentioned above is that the representation
partially covers the modelling methods. That is, complete model-
ð6Þ ling with formal languages is not feasible for all methods. Thus,
two levels of representation for ID methods were defined: full
and partial. The category unrepresented had also to be added. This
category identifies those methods for which no practical rule-
based representation has been found to date. Table 3 shows some
examples of these three categories.
Some partially represented methods use ambiguous qualifiers
such as ‘‘moderately complex’’ or ‘‘enough’’. In the current represen-
tation, conventions are used to model this kind of ambiguity. For
example, a moderately complex concept corresponds to those that
have more than two deep level in relation to the concepts they in-
clude or to those which include more than one concept, at least one
of which has at least one deep level. Currently, SWRL language
does not support ambiguity in representation. However, in future
extensions fuzzy representations could be incorporated to repre-
sent ambiguous categories, and thus serve to improve the cata-
logue [31,46].
The common rules are used to represent methods that have been
found to be applicable in the context of different theories. For exam-
ple, the method of Learning by Doing theory that says ‘‘Must have
decision points with consequences that become evident’’ could share
a rule with the method ‘‘Give students some choice as to which versions
of the task to learn next’’ from Elaboration theory.
In this case, after considering a partial representation of both
methods, the following rule might be common to both:
ld : Methodð?mÞ ^ ld : Playð?pÞ ^ ld : play  refð?m;?pÞ ^ ld
: act  refð?p;?acÞ ^ ld : Actð?acÞ ^ ld : Role  Partð?rpÞ^
ld : role  part  refð?ac; ?rpÞ ^ ld : LearningActivity
ð?laÞ ^ ld : execution  entity  refð?rp; ?laÞ ^ ld
: On  Completion  Unitð?ocÞ ^ ld : on  completion
 refð?la; ?ocÞ ^ ld : complete  unit  of  learning
ð?oc;user  choiceÞ ! rr : userChoice  LActivityrule
ð?m;“true”Þ
According to the structure of IMS LD, a method contains one or
more plays and they have one or more acts. The role-part structure
links an act with both a role and learning activity. The learning activ-
ities are referenced by LD: execution-entity-ref relation. By
property ld: on-completion-ref it is possible to define how
the LD behaves after each learning activity.
The previous rule states that at the end of an activity, the learner
can choose the next activity to perform. This rule enables the scope
for decision to be modelled in forthcoming student learning activities.
There are also rules that are used in different method represen-
tations of the same theory. For example, consider the method from
Elaboration Theory: ‘‘Group concepts and their supporting content
into learning episodes that are not so large as to make review and
synthesis difficult but are not so small as to break up the flow of the
learning process’’. In general, it has some similarity with the method
‘‘Groupsteps/ principles and their supporting content into learning epi-
sodes’’. Although the first method may require other additional
rules, the following rule may be applied to both:
ld : Methodð?mÞ ^ ld : Playð?pÞ ^ ld : play  refð?m; ?pÞ ^ ld
: act  refð?p; ?ac1Þ ^ ld : act  refð?p; ?ac2Þ ^ ld
: Actð?ac1Þ ^ ld : Actð?ac2Þ ^ differentFromð?ac1; ?ac2Þ
! rr : has2actsð?m; “true”Þ
186 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194

Table 3
Example of methods with complete and partial representation.

ID theory Method/submethod SWRL rule Observation

Elaboration Give students some choice as
to which concepts to elaborate
upon first/next
ld:Method(?m) ^ Fully represented. Both rules can
ld:Play(?p) ^ represent the directions of the
ld:play-ref(?m, ?p) ^ method in the sense that
ld:On-Completion-Unit(?oc) ^ students can choose the next
ld:on-completion-ref(?p, ?oc) ^ play and act of LD
complete-unit-of-learning(?oc, "user-choice")
? rr:userChoice-play(?m, "true")
ld:Method(?m) ^
ld:Play(?p) ^
ld:play-ref(?m, ?p) ^
ld:Act(?ac1) ^
ld:act-ref(?p, ?ac1) ^
ld:On-Completion-Unit(?oc) ^
ld:on-completion-ref(?ac1, ?oc) ^
complete-unit-of-learning(?oc, "user-choice")
? rr:userChoice-act(?m, "true")

Learning by Doing Must allow enough
opportunities to practice the
skill and seek the knowledge
uol:Learning-Object (?lo) ^ Partially represented. It identifies
uol:Metadata (?m) ^ several types of learning objects
uol:metadata-ref (?lo, ?m) ^ that allow skills to be practiced
lom:LOM (?lom) ^ skills and the knowledge to be
uol:metadata-description (?m, ?lom) ^ sought. However, the ambiguity
lom:Educational (?edu) ^ of the qualifier ‘‘enough’’ makes
lom:educational-ref (?lom, ?edu) ^ representation difficult
lom:learningresourcetype (?lo, "exer")?
practiceSkill (?lo, "true")
uol:Learning-Object (?lo) ^
uol:Metadata (?m) ^
uol:metadata-ref (?lo, ?m) ^
lom:LOM (?lom) ^
uol:metadata-description (?m, ?lom) ^
lom:Educational (?edu) ^
lom:educational-ref (?lom, ?edu) ^
lom:learningresourcetype (?lo, "simulation")
? practiceSkill (?lo, "true")

Multiple Intelligences Resists the temptation to None Unrepresented. This theory

represent the topic in one generated multiple forms of
‘‘optimal’’ mode content representation thus
making it difficult to know
which was optimal

Assuming a certain similarity between the concept ‘‘learning
episode’’ and an act of an LD, this rule verifies that the LD has at
least two acts in its structure.
5.2. Elaboration theory modelling
This section shows in greater detail the representation of Elab-
oration theory. Elaboration theory has three main methods: (1)
Conceptual Elaboration Sequence, (2) Theoretical Elaboration Se-
quence, and (3) Simplifying Conditions Sequence. Table 4 shows
a representation fragment of this theory. Methods, sub-methods,
identification of the rule in the catalogue, and an observation about
the modelling are shown.
Some of the rules presented can be used to model the guides for
multiple methods, either within a same theory or a different one.
This is the case of the rule called rr:showsBefore that from
two activities grouped in a structure of activities indicates the exe-
cution order. Similarly, rule rr:userChoice-Lactivity can be
used in different theories and thus provide students with the
opportunity to choose their learning paths.
As stated, not all methods are fully representable using this ap-
proach. This is because in some cases they have to do with con-
cepts which are not not considered in the scope of IMS LD. To be
more precise, some concepts used in the DI methods have little
or no relation to learning activities or resources, and this impedes
representation using this approach. Their representation requires
extra knowledge about the learning process. For example, consider
the method ‘‘For procedural task focus on teaching steps; for heuris-
tics task focus on teaching principles; and for combination task teach
both steps and principles – in accordance with the way experts thing
about the task’’. In this case, IMS LD standard does not differentiate
explicitly between procedural task or heuristics task, so that the
representation of this method is beyond the scope of this approach.
6. A case study
To demonstrate the usefulness of the proposed representation
model, we present a case study that shows a fragment of a LD
for an introductory course in cell biology. The course’s learning
objectives have to do with learning how cells grow and divide,
and how chromosomes are split up during mitosis and meiosis. It
has been designed for biology students at the college level, but is
useful for all types of people interested in the topic. The course
requires prior basic knowledge of cell biology. Course contents
are based on those represented in Gene Ontology [10]. This domain
ontology provides a controlled terminology and a formal structure
for describing concepts and semantic annotations. It is supported
by the GO Consortium and the National Human Genome Research
Institute (NHGRI).

Table 4
Fragment of modelling of Reigeluth’s Elaboration theory.
Guide
Submethod 1. Conceptual Elaboration Sequence
1 Use this approach when the goals call Met1-1-1
for learning many related concepts
2 Teach broader, more inclusive
concepts before the narrower, more
detailed concepts that elaborate upon
them
C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194 187
Id rules SWRL rules Observations
KnowledgeItem (?c1) ^ – If a concept includes two other
KnowledgeItem (?c2) ^ concepts, it is considered to have
KnowledgeItem (?c3) ^ many related concepts
concept-includes (?c1, ?c2) ^
concept-includes (?c1, ?c3) ^
differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
ConceptLearningActivity (?la) ^
concept-learning-objective (?la, ?c1)
? rr:hasRelatedConcepts-includes (?la, "true")
Met1-1-2 KnowledgeItem (?c1) ^ – If a concept is related to two other
KnowledgeItem (?c2) ^ KnowledgeItem (?c3) concepts, by a hasKind relation, it is
concept-hasKind (?c1, ?c2) ^ considered to have many related
concept-hasKind (?c1, ?c3) ^ concepts
differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
ConceptLearningActivity (?la) ^
concept-learning-objective (?la, ?c1)
? rr:hasRelatedConcepts-hasKind (?la, "true")
Met1-1-3 KnowledgeItem (?c1) ^ – If a concept is related to two other
KnowledgeItem (?c2) ^ concepts, by a hasPart relation, it is
KnowledgeItem (?c3) ^ considered to have many related
concept-hasPart (?c1, ?c2) ^ concepts
concept-hasPart (?c1, ?c3) ^
differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
ConceptLearningActivity (?la) ^
concept-learning-objective (?la, ?c1)
? rr:hasRelatedConcepts-hasPart (?la, "true")
Met1-1-4 KnowledgeItem (?c1) ^ KnowledgeItem (?c2) ^ Knowl- – If a concept is related to two other
edgeItem (?c3) ^ concepts, by an hasSupportingContent
concept-hasSupportingContent (?c1, ?c2) relation, it is considered to have
^ concept-hasSupportingContent (?c1, ?c3) many related concepts
^ differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
ConceptLearningActivity (?la) ^
concept-learning-objective (?la, ?c1)
? rr:hasRelatedConcepts-
supportingContent (?la, "true")
showsBefore Ld:Learning-Activity (?a1) ^ Rule that determines which activity
ld:Learning-Activity (?a2) ^ are executed before another
differentFrom (?a1, ?a2) ^
ld:Activity-Structure (?as1) ^
ld:execution-order (?a1, ?o1) ^
ld:execution-order (?a2, ?o2) ^
ld:execution-entity-ref (?as1, ?a1) ^
ld:execution-entity-ref (?as1, ?a2) ^
swrlb:lessThan (?o1, ?o2)
? rr:showsBefore (?a2, ?a1)
Met1-2 RET-showsBefore (?a1, ?a2) ^ Checks that activities associated with
ConceptLearningActivity (?a1) ^ concepts ‘‘broader’’ are run first. If one
ConceptLearningActivity (?a2) ^ concept includes another, it is
differentFrom (?a1, ?a2) ^ regarded as broader
ld:Activity-Structure (?as) ^
ld:execution-entity-ref (?as, ?a1) ^
ld:execution-entity-ref (?as, ?a2) ^
concept-learning-objective (?a1, ?c1) ^
concept-learning-objective (?a2, ?c2) ^
KnowledgeItem (?c1) ^
KnowledgeItem (?c2) ^
concept-includes (?c2, ?c1) ^
differentFrom (?c1, ?c2)
? rr:broaderFirst (?as, "true")
(continued on next page)
188
Table 4 (continued)
Guide
4 Teach ‘‘supporting’’ content
(principles, procedures, information,
higher-order thinking skills,
attitudes, etc.) together with the
concepts to which they are most
closely related
5 Group concepts and their supporting
content into ‘‘learning episodes’’ that
are not so large as to make review
and synthesis difficult but are not so
small as to break up the flow of the
learning process
6 Give students some choice as to
which concepts to elaborate upon
first/next
Submethod 3. Simplifying conditions sequence
1 Use this approach when the goals call Met3-1-1
for learning a task at least moderate
complexity
5 Teach supporting content together
with the steps or/and principles to
which they are more closely related
C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
Id rules SWRL rules Observations
Met1-4 KnowledgeItem (?k) ^ Checks that a concept has an
KnowledgeItem (?k1) ^ associated SupportingContent
concept-hasSupportingContent (?k, ?k1) ^
ConceptLearningActivity (?la) ^
concept-learning-objective (?la, ?k)
? rr:hasSupportingContent (?la, true)
Met1-5 ld:Method (?m) ^ Checks that the method has a play
ld:Play (?p) ^ and at least two acts
ld:play-ref (?m, ?p) ^
ld:act-ref (?p, ?ac1) ^
ld:act-ref (?p, ?ac2) ^
ld:Act (?ac1) ^
ld:Act (?ac2) ^
differentFrom (?ac1, ?ac2)
? rr:has2acts (?m, "true")
Met1-6-1 ld:Method (?m) ^ Checks that the learner may choose
ld:Play (?p) ^ the next play
ld:play-ref (?m, ?p) ^
ld:On-Completion-Unit (?oc) ^
ld:On-Completion-ref (?p, ?oc) ^
complete-unit-of-learning (?oc,"user-choice")
? rr:userChoice-play (?m, "true")
Met1-6-2 ld:Method (?m) ^ Checks that the learner may choose
ld:Play (?p) ^ the next act
ld:play-ref (?m, ?p) ^
ld:Act (?ac1) ^
ld:act-ref (?p, ?ac1) ^
ld:On-Completion-Unit (?oc) ^
ld:on-completion-ref (?ac1, ?oc) ^
complete-unit-of-learning (?oc, "user-choice")
? rr:userChoice-act (?m, "true")
KnowledgeItem (?c1) ^ KnowledgeItem (?c2) ^ Knowl- Verifies that moderately complex
edgeItem (?c3) ^ KnowledgeItem (?c4) ^ concepts are used. A concept is
differentFrom (?c1, ?c2) ^ moderately complex if it has more
differentFrom (?c1, ?c3) ^ than two deep levels in relation to the
differentFrom (?c1, ?c4) ^ concepts that includes. Checks that
differentFrom (?c2, ?c3) ^ are used this concepts
differentFrom (?c2, ?c4) ^
differentFrom (?c3, ?c4) ^
concept-includes (?c1, ?c2) ^
concept-includes (?c2, ?c3) ^
concept-includes (?c3, ?c4) ^
ConceptLearningActivity (?cla) ^
concept-learning-objective (?cla, ?c1)
? rr:moderatlyComplex (?cla, "true")
Met3-1-2 KnowledgeItem (?c1) ^ KnowledgeItem (?c2) ^ Knowl- Verifies that moderately complex
edgeItem (?c3) ^ KnowledgeItem (?c4) ^ concepts are used. A concept is also
differentFrom (?c1, ?c2) ^ considered as moderately complex if
differentFrom (?c1, ?c3) ^ it includes more than one concept
differentFrom (?c1, ?c4) ^ and one of t hem has at least one
differentFrom (?c2, ?c3) ^ deep level
differentFrom (?c2, ?c4) ^
differentFrom (?c3, ?c4) ^
concept-includes (?c1, ?c2) ^
concept-includes (?c1, ?c3) ^
concept-includes (?c3, ?c4) ^
ConceptLearningActivity (?cla) ^
concept-learning-objective (?cla, ?c1)
? rr:moderatlyComplex (?cla, "true")
Met3-5-1 ld:Method (?m) ^ Checks that the method teaches a
ld:Play (?p) ^ concept and content-support in a
ld:play-ref (?m, ?p) ^ single play
ld:act-ref (?p, ?ac) ^
ld:Act (?ac1) ^
ld:Act (?ac2) ^
ld:Role-Part (?rp1) ^
ld:Role-Part (?rp2) ^
differentFrom (?rp1, ?rp2) ^
ld:role-part-ref (?ac1, ?rp1) ^
 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194 189

Table 4 (continued)

Guide Id rules SWRL rules Observations

ld:role-part-ref (?ac2, ?rp2) ^
ConceptLearningActivity (?la1) ^
ld:execution-entity-ref (?rp1, ?la1) ^
ConceptLearningActivity (?la2) ^
ld:execution-entity-ref (?rp2, ?la2) ^
differentFrom (?la1, ?la2) ^
concept-learning-objective (?la1, ?c1) ^ concept-
learning-objective (?la2, ?c2) ^ KnowledgeItem
(?c1) ^
KnowledgeItem (?c2) ^
differentFrom (?c1, ?c2) ^
concept-hasSupportingContent (?c1, ?c2)
? rr:SupportingContent (?m, "true")
Met3-5-2 ld:Method (?m) ^ ld:Play (?p) ^ Checks that the method teaches a
ld:play-ref (?m, ?p) ^ concept and content-support of
ld:act-ref (?p, ?ac) ^ another that contains it, in the same
ld:Act (?a1) ^ ld:Act (?ac2) ^ play
ld:Role-Part (?rp1) ^
ld:Role-Part (?rp2) ^
ld:role-part-ref (?ac1, ?rp1) ^
ld:role-part-ref (?ac2, ?rp2) ^
differentFrom (?rp1, ?rp2) ^
ConceptLearningActivity (?la1) ^
ld:execution-entity-ref (?rp1, ?la1) ^
ConceptLearningActivity (?la2) ^
ld:execution-entity-ref (?rp2, ?la2) ^
differentFrom (?la1, ?la2) ^
concept-learning-objective (?la1, ?c2) ^
concept-learning-objective (?la2, ?c3) ^
KnowledgeItem (?c1) ^
KnowledgeItem (?c2) ^
KnowledgeItem (?c3) ^
differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
concept-includes (?c1, ?c2) ^
concept-hasSupportingContent (?c1, ?c3)
? rr: SupportingContent (?m, "true")
Met3-5-3 ld:Method (?m) ^ ld:Play (?p) ^ Checks that the method includes, in
ld:play-ref (?m, ?p) ^ one play, a concept and content-
ld:act-ref (?p, ?ac) ^ support of another concept, which is
ld:Act (?ac1) ^ ld:Act (?ac2) ^ a part
ld:Role-Part (?rp1) ^ ld:Role-Part (?rp2)
^ ld:role-part-ref (?ac1, ?rp1) ^
ld:role-part-ref (?ac2, ?rp2) ^
differentFrom (?rp1, ?rp2) ^
ConceptLearningActivity (?la1) ^
ld:execution-entity-ref (?rp1, ?la1) ^
ConceptLearningActivity (?la2) ^
ld:execution-entity-ref (?rp2, ?la2) ^
differentFrom (?la1, ?la2) ^
concept-learning-objective (?la1, ?c2) ^
concept-learning-objective (?la2, ?c3) ^
KnowledgeItem (?c1) ^ KnowledgeItem (?c2)
^ KnowledgeItem (?c3) ^
differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
concept-hasPart (?c1, ?c2) ^
concept-hasSupportingContent (?c1, ?c3)
? rr:SupportingContent (?m, "true")
Met3-5-4 ld:Method (?m) ^ ld:Play (?p) ^ Checks that the method includes, in
ld:play-ref (?m, ?p) ^ one play, a concept and content-
ld:act-ref (?p, ?ac) ^ support of another concept which is
ld:Act (?ac1) ^ ld:Act (?ac2) ^ ‘‘a kind’’
ld:Role-Part (?rp1) ^
ld:Role-Part (?rp2) ^
ld:role-part-ref (?ac1, ?rp1) ^
ld:role-part-ref (?ac2, ?rp2) ^
differentFrom (?rp1, ?rp2) ^
ConceptLearningActivity (?la1) ^
ld:execution-entity-ref (?rp1, ?la1) ^
ConceptLearningActivity (?la2) ^
ld:execution-entity-ref (?rp2, ?la2) ^

(continued on next page)

190 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194

Table 4 (continued)

Guide Id rules SWRL rules Observations

differentFrom (?la1, ?la2) ^
concept-learning-objective (?la1, ?c2) ^
concept-learning-objective (?la2, ?c3) ^
KnowledgeItem (?c1) ^
KnowledgeItem (?c2) ^
KnowledgeItem (?c3) ^
differentFrom (?c1, ?c2) ^
differentFrom (?c1, ?c3) ^
differentFrom (?c2, ?c3) ^
concept-hasKind (?c1, ?c2) ^
concept-hasSupportingContent (?c1, ?c3)
? rr:SupportingContent (?m, "true")

6 Group steps/principles and their
supporting content into ‘‘learning
episodes’’
Met3-6-1 ld:Method (?m) ^ ld:Play (?p) ^ Verifies that a method has a play and
ld:play-ref (?m, ?p) ^ at least two acts
ld:act-ref (?p, ?ac1) ^
ld:act-ref (?p, ?ac2) ^
ld:Act (?ac1) ^ ld:Act (?ac2) ^
differentFrom (?ac1, ?ac2)
? rr:has2acts (?m, "true")
Met3-6-2 ld:Method (?m) ^ ld:Play (?p) ^ Checks that the method includes an
ld:play-ref (?m, ?p) ^ Activity-Structure, which links at least
ld:act-ref (?p, ?ac) ^ two activities
ld:Act (?ac) ^ ld:Role-Part (?rp) ^
ld:role-part-ref (?ac, ?rp1) ^
ld:Activity-Structure (?sa) ^
ld:execution-entity-ref (?rp1, ?sa) ^
ConceptLearningActivity (?la1) ^
ConceptLearningActivity (?la2) ^
differentFrom (?la1, ?la2) ^
ld:execution-entity-ref (?sa, ?la1) ^
ld:execution-entity-ref (?sa, ?la2) ^
?rr:hasActivityStructure (?m, "true")

7 Give students some choice as to
which versions of the task to learn
next
Met3-7 ld:Method (?m) ^ ld:Play (?p) ^ Checks that the learner may choose
ld:play-ref (?m, ?p) ^ the next learning activity
ld:act-ref (?p, ?ac) ^
ld:Act (?ac) ^ ld:Role-Part (?rp) ^
ld:role-part-ref (?ac, ?rp) ^
ConceptLearningActivity (?la) ^
ld:execution-entity-ref (?rp, ?la) ^
ld:On-Completion-Unit (?oc) ^
ld:on-completion-ref (?la, ?oc) ^
complete-unit-of-learning (?oc, "user-choice")
? rr:userChoice-Lactivity (?m, "true")

The application of the Elaboration theory requires the building used in the LD and shows is-a and part-of kind of relationships used
of a conceptual structure containing the concepts that cover the in the ontology. These properties are formally defined in the GO
course. In this case study, the concepts are related to cell cycle pro- and provide an extensive and detailed account of the domain, used
cess. Fig. 2 is a fragment of the GO ontology that contains concepts regularly by researchers in bioinformatics.

Fig. 2. Concepts used in the LD from GO ontology. Figure generated with OBO-Edit ontology editor available in http://www.oboedit.org/.
 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194 191

To select the topics to be incorporated in the design, Elaboration

theory proposes the use of two options: topical and spiral sequenc-
ing. In this case study, the subjects were selected using topical
sequencing [36] in order to simplify selection. According to the
IMS-LD specification, LD built as a case study consists of a play
and the three following acts: overview of cellular process, inter-
phase, and mitosis. In learning activities involving the roles ‘‘tea-
cher’’ and ‘‘student’’, the learning objectives defined for the LD are
related to learning of growth and cell division, mitosis and meiosis
process. Basic knowledge of cell biology was defined as a prerequi-
site. With respect to content, the LD uses GO ontology to cover top-
ics such as cellular process, mitosis and meiosis. A partial view of
the resulting LD resulting is shown in Fig. 3 using Recourse design
tool [4].
The design tools compatible with IMS-LD generate a XML file
called imsmanifest.xml, which contains the information associ-
ated with the structure and used resource by LD. Fig. 4 shows an
abbreviated portion of XML code that represents the LD. Fig. 5. Instances of classes Method, Play and Act of IMS-LD ontology.

Fig. 3. Overview of LD in recourse tool.

Fig. 4. Fragment of imsmanifest.xml generated for the LD.

192 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194

Fig. 6. A learning object used in the construction of LD. Retrieved from MERLOT repository (http://www.biology.arizona.edu/cell_bio/cell_bio.html).

In the XML the proposed LD structure, containing a method, a
play and three acts, can be observed. One of these acts is shown
in detail. The information contained in imsmanifest.xml is
instantiated into IMS-LD ontology.
Fig. 5 shows the instances for classes ld:Method, ld:Play and
ld:Act, which correspond to one method, play and events
respectively.
Some learning objects from MERLOT repository [27], were used
in the LD construction. Fig. 6 shows part of an object called ‘‘The
Biology Project: Cell Biology’’, which meets the objectives of the
course. This figure also shows the high evaluation given the re-
source by repository users in relation to quality content, effective-
ness and easy use.

Fig. 7. KnowledgeItem class instances of IMS-LD ontology.

 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
Table 5
Some of the rules satisfied by the LD.
Id rule LD description
rr:hasRelatedConcepts- This rule validates that objectives considered
includes in the LD have many related concepts. See
KnowledgeItem class in Fig. 6
rr:has2acts The LD has a play and more than one act, so
the contents are considered to be grouped into
‘‘learning episodes’’. See ld:Play and ld:Act
classes in Fig. 5
rr:moderatlyComplex The contents of LD are ‘‘moderately complex’’,
since it considers a concept that includes at
least two sub-concepts, and one of them has
more than one deep level. See
KnowledgeItem class in Fig. 6
In order to clarify the example, instances of the class
tgs:KnowledgeItem (described in an example above) are shown.
This class is used to represent the concepts covered by the LD (see
Fig. 7). This way, the GO is mapped to the generic knowledge item
scheme described above. This mapping can be done similarly with
any other existing domain ontology, thus enabling reuse.
Once LD information is instantiated in the ontology, it is possi-
ble to check if the LD follows the rules representing the guides for
Elaboration theory. Table 5 shows some of the rules covered by the
LD, for the case study presented.
Moreover, the rule rr:userChoice-Lactivity is not satis-
fied, since it verifies that the student can choose the next learning
activity, something which did not happen in the case-study.
Although the case-study showed that three of the model’s rules
were satisfied, according to the representation proposed in this re-
search, all rules must be covered.
7. Conclusions and outlook
According to Reigeluth, instructional design theories offer expli-
cit guidance on how to help people learn. These theories, which
may be of use in the construction of learning resources, are cur-
rently expressed in natural language. Their translation to lan-
guages with computational semantics yields the benefit of
enabling the construction of theory-aware tools to support the
instructional design process. This paper has described a framework
for the representation of instructional design methods, modelled as
a collection of methods using OWL and SWRL language.
The approach used for modelling was based on a IMS LD ontol-
ogy that represents the structure and sequence of learning activi-
ties, participants, resources and services included in a learning
design. To date, using the concept of legal interpretation, formal
representations have been proposed to partially cover the methods
of three theories: Elaboration theory, Multiple Intelligences and
Learning by Doing. However, the catalogue is constantly open to
improvement and the incorporation of new representations of the-
ories. Full results of modelling are available at http://www.ier-
u.org/idont-wiki/.
Some ID methods cannot be represented because they use con-
cepts that are beyond computational representation. Other meth-
ods may require the use of fuzzy modelling for the categorisation
of concepts. Yet other methods require extra knowledge about
the learning process—for example, the learning styles of stu-
dents—for their representation.
The representation presented here gives rise to what we call a
rule catalogue. The use of the standard IMS-LD allows the cata-
logue to become a knowledge base for any application that brings
together knowledge in a format compatible with this standard.
Moreover, the use of formal ontologies allows the represented
knowledge to be shared and used by various applications. For
193
example, it is possible to develop applications that attempt to
check the conformity of a LD with a particular legal interpretation
of an instructional design theory. This would develop a wizard for
applying these theories in order to inform and guide the user
through the design of a learning design in authoring tools.
Furthermore, the model can be used to generate a skeleton for a
given learning design theory, which would prove useful to inexpe-
rienced users in the application of one instructional design theory.
Another use of the proposed model is to provide support to the
search and retrieval of LD that bring together certain methods of
an instructional theory.
Future work related to this research should be in two direc-
tions: the continuous revision of the representations contained in
the catalogue, and the construction of tools to assist in the imple-
mentation of instructional design methods.
In relation to the first point, research should be aimed at incor-
porating new representations of theories and improving existing
ones by considering the possible updating of formal ontological
language. As for the second, efforts should be focused on the con-
struction of a prototype to evaluate the conformance of an LD with
a particular IDT. This prototype would evaluate the usefulness of
the models in a design environment. Using theory representations
contained in the catalogue presented here, the prototype could re-
port on aspects of LD that violate the guidelines of the ID methods,
as well as guiding guide the designer in design improvements
which might achieve conformance with a IDT. This tool should
be integrated into any learning design tool compatible with IMS-
LD specification.
Acknowledgments
This work was partially supported by DIUBB 125515 3/RS Pro-
ject of University of Bio-Bio and MECESUP UBB 0305 Project, Chile;
by Project MAVSEL (Ref. TIN2010-21715-C02-01) funded by the
Spanish Ministry of Science and Innovation, Spain; and the Projects
SCAIWEB2 (PEIC09-0196-3018) and PLINIO (POII10-0133-3516) of
the Autonomous Government of Castilla-La Mancha, Spain.
References
[1] G. Albano, M. Gaeta, S. Salerno, E-learning: a model and process proposal,
International Journal of Knowledge and Learning 2 (2006) 73–88.
[2] R. Amorim, M. Lama, E. Sánchez, A. Riera, X. Vila, A learning design ontology
based on the IMS specification, Educational Technology and Society 9 (2006)
38–57.
[3] R. Amorim, E. Sánchez, M. Lama, S. Barro, X. Vila, Representación y ejecución de
unidades educativas a través de una ontologı´a de diseños de aprendizaje,
Revista Iberoamericana de Inteligencia Artificial 33 (2007) 61–77.
[4] P. Beauvoir, D. Griffiths, P. Sharples, Learning design authoring tools in the
TENCompetence project, in: R. Koper (Ed.), Learning Network Services for
Professional Development, 2009, pp. 379–387
[5] T. Berners-Lee, J. Hendler, O. Lassila, The Semantic Web, Scientific American,
Publishing, 2001. pp. 28–37.
[6] I. Bittencourt, E. Costa, M. Silva, E. Soares, A computational model for
developing semantic web-based educational systems, Knowledge-Based
Systems 22 (2009) 302–315.
[7] H. Boley, S. Tabet, G. Wagner, Design rationale for RuleML: a markup language
for semantic web rules, in: Proceeding of International Conference on Semantic
Web and Web Services, 2001, pp. 381–401.
[8] J. Bourdeau, R. Mizoguchi, Y. Hayashi, V. Psyche, R. Nkambou, When the
domain of the ontology is education, in: Proc. of the 4th Conf. on Intelligent,
Interactive Learning Objects Repository Networks (I2LOR’07), 2007.
[9] W. Dick, L. Carey, J. Carey, Systematic Design of Instruction, seventh ed., Allyn
& Bacon, Inc., 2008.
[10] A. Diehl, J. Lee, R. Scheuermann, J. Blake, Ontology development for biological
systems: immunology, Bioinformatics 23 (2007) 913–915.
[11] S. Downes, Learning objects: resources for distance education worldwide, The
International Review of Research in Open and Distance Learning 2 (2001).
[12] R. Gagne, L. Briggs, W. Wager, Principles of Instructional Design, fourth ed.,
Wadsworth Pub., 1992.
[13] H. Gardner, Multiple approaches to understanding, in: C. Reigeluth (Ed.),
Instructional-Design Theories and Models. A New Paradigm of Instructional
Theory, vol. II, 1999.
194 C. Vidal-Castro et al. / Knowledge-Based Systems 33 (2012) 180–194
[14] D. Griffiths, J. Blat, R. Garcı́a, H. Vogten, K. Kwong, Learning design tools, in: R.
Koper, C. Tattersall (Eds.), A Handbook on Modelling and Delivering
Networked Education and Training, New York, 2005.
[15] B. Harper, S. Agostinho, S. Bennett, J. Lukasiak, L. Lockyer, Constructing high
quality learning environments using learning designs and learning objects, in:
Fifth IEEE International Conference on Advanced Learning Technologies
(ICALT’05), 2005, pp. 266–270.
[16] Y. Hayashi, J. Bourdeau, R. Mizoguchi, Ontological support for a theory-eclectic
approach to instructional and learning design, in: W. Nejdl, K. Tochtermann
(Ed.), EC-TEL 2006, Lecture Notes on Computer Science 4227, 2006, pp. 155–
169.
[17] R. Heinich, M. Molenda, D. Russell, Instructional Media and the New
Technologies of Instruction, Macmillan Pub. Co., New York, 1993.
[18] M. Hogo, Evaluation of e-learning systems based on fuzzy clustering models
and statistical tools, Expert Systems with Applications 37 (2010) 6891–6903.
[19] I. Horrocks, P. Patel-Schneider, D. McGuinness, C. Welty, OWL: a description
logic based ontology language for the semantic web, in: F. Baader, D.
Calvanese, D. McGuinness, D. Nardi, P. Patel-Schneider (Eds.), The
Description Logic Handbook: Theory, Implementation, and Applications,
second ed., 2007.
[20] J. Keller, Development and use of the ARCS model of motivational design,
Journal of Instructional Development 10 (1987) 2–10.
[21] C. Knight, D. Gašević, G. Richards, An ontology-based framework for bridging
learning design and learning content, Educational Technology and Society 9
(2006) 23–37.
[22] R. Koper, Modeling Units of Study From a Pedagogical Perspective: the
Pedagogical Meta-model Behind EML, Educational Technology Expertise
Centre, Open University of the Netherlands, 2001.
[23] R. Koper, Educational Modelling Language: Adding Instructional Design to
Existing Specifications, 2002.
[24] R. Koper, Y. Miao, Using the IMS LD standard to describe learning designs, in: L.
Lockyer, S. Bennet, S. Agostinho, B. Harper (Eds.), Handbook of Research on
Learning Design and Learning Objects: Issues, Applications and Technologies,
2008, pp. 41–86.
[25] R. Koper, C. Tattersall, Learning Design: A Handbook on Modelling and
Delivering Networked Education and Training, Springer Verlag, Heidelberg,
2005.
[26] M. Lama, R. Amorim, E. Sánchez, X. Vila, Semantic description of the IMS learning
design, in: The 12th International Conference on Artificial Intelligence in
Education (AIED 2005), Workshop Applications of Semantic Web Technologies
for E-Learning (SW-EL’05), Amsterdam, Holanda, 2005, pp. 37–46.
[27] F. McMartin, MERLOT: a model for user involvement in digital library design
and implementation, Journal of Digital Information 5 (2004).
[28] M.D. Merril, Instructional transaction theory: an instructional design model
based on knowledge objects, Educational Technology 36 (1996) 7.
[29] R. Mizoguchi, J. Bourdeau, Theory-aware authoring environment-ontological
engineering approach, in: Proceedings International Conference on Computers
in Education (ICCE’02), 2002.
[30] M. O’Connor, R. Shankar, S. Tu, C. Nyulas, A. Das, in: Developing a Web-Based
Application using OWL and SWRL AAAI Spring Symposium, Stanford, CA, USA,
2008.
[31] J. Pan, G. Stoilos, G. Stamou, V. Tzouvaras, I. Horrocks, f-SWRL: a fuzzy
extension of SWRL, Journal on Data Semantics VI (2006) 28–46. special issue
on Emergent semantics.
[32] G. Paquette, Foundations of instructional engineering, in: G. Paquette (Ed.),
Instructional Engineering in Networked Environments, 2004.
[33] G. Paquette, K. Lundgren-Cayrol, M. Léonard, The MOT+ visual language for
knowledge-based instructional design, in: L. Botturi, T. Stubs (Eds.), Handbook
on Virtual Instructional Design Languages, 2008.
[34] C. Peterson, Bringing ADDIE to life: instructional design at its best, Journal of
Educational Multimedia and Hypermedia 12 (2003) 227–241.
[35] A. Rawlings, P. van Rosmalen, R. Koper, M. Rodrı́gues-Artacho, P. Lefrere,
Survey of Educational Modelling Languages (EMLs), Version 1, CENN/ISSS,
2002.
[36] C.M. Reigeluth, Instructional-Design Theories and Models, A New Paradigm of
Instructional Theory, vol. II, Lawrence Erlbaum Assoc., Mahwah, NJ, 1999.
[37] M. Rodrı́guez, M. Serra, J. Cabot, I. Guitart, Evolution of the teacher roles and
figures in e-learning environments, in: Proceedings of the Sixth IEEE
International Conference on Advanced Learning Technologies (ICALT’06), July
05–07, 2006, pp. 512–514.
[38] C. Romero, S. Ventura, Educational data mining: a survey from 1995 to 2005,
Expert Systems with Applications 33 (2007) 135–146.
[39] E. Sanchez, M. Lama, Técnicas de Inteligencia Artificial Aplicadas a la
Educación, Revista Iberoamericana de Inteligencia Artificial 33 (2007) 7–12.
[40] R. Schank, T. Berman, K. Macpherson, Learning by doing, in: C. Reigeluth (Ed.),
Instructional-Design Theories and Models. A New Paradigm of Instructional
Theory, vol. II, 1999.
[41] M.A. Sicilia, Beyond content: sharing the design of open educational resources,
Revista de Universidad y Sociedad del Conocimiento (RUSC) 4 (2007).
[42] M.A. Sicilia, On the general structure of ontologies of instructional models, in:
Proceedings del IV Simposio Pluridisciplinar sobre Diseño, Evaluación y
Desarrollo de Contenidos Educativos Reutilizables, SPDECE 2007, Bilbao,
Spain, 2007b.
[43] M.A. Sicilia, M. Lytras, On the representation of change according to different
ontologies of learning, International journal of learning and change 1 (2005)
66–79.
[44] M.A. Sicilia, M. Lytras, S. Sánchez-Alonso, E. Garcı́a-Barriocanal, M. Zapata-Ros,
Modeling instructional-design theories with ontologies: using methods to
check, generate and search learning designs, Computer in Human Behavior 27
(2011) 1389–1398.
[45] M.A. Sicilia, S. Sánchez-Alonso, E. Garcı́a-Barriocanal, On supporting the
process of learning design through planners, in: Proceeding of the Virtual
Campus Workshop, Barcelona, 2006, pp. 81–89.
[46] Z. Wang, Z. Ma, Y., Yan, X. Meng, Vague-SWRL: a fuzzy extension of SWRL, in:
Proceedings of the 2nd International Conference on Web Reasoning and Rule
Systems, Lecture Notes in Computer Science, vol. 5341, Karlsruhe, Germany,
2008.
[47] A. Zouaq, R. Nkambou, C. Frasson, An integrated approach for automatic
aggregation of learning knowledge objects, Interdisciplinary Journal of
Knowledge and Learning Objects (IJKLO) 3 (2007) 135–162.