Content uploaded by Dietrich Albert
Author content
All content in this area was uploaded by Dietrich Albert
Content may be subject to copyright.
Competence-Based Knowledge Structures for
Personalised Learning
JÜRGEN HELLER, CHRISTINA STEINER, CORD HOCKEMEYER,
AND DIETRICH ALBERT
University of Graz - Graz, Austria
juergen.heller@uni-graz.at
chr.steiner@uni-graz.at
cord.hockemeyer@uni-graz.at
dietrich.albert@uni-graz.at
Competence-based extensions of Knowledge Space Theory
are suggested as a formal framework for implementing key
features of personalised learning in technology-enhanced
learning. The approach links learning objects and assessment
problems to the relevant skills that are taught or required. Var-
ious ways to derive these skills from domain ontologies are
discussed in detail. Moreover, it is shown that the approach
induces structures on the assessment problems and learning
objects, respectively, that can serve as a basis for an efficient
adaptive assessment of the learners’ skills, and for selecting
personalised learning paths.
Personalised learning aims to tailor teaching to individual needs, inter-
ests, and aptitude to ensure that every learner achieves and reaches the high-
est standards possible. It usually proceeds by assessing the learner’s current
knowledge state and probably other individual characteristics or prefer-
ences, and by using the results of this assessment to inform further teaching.
Knowledge Space Theory (Doignon & Falmagne, 1985, 1999; Falmagne,
Koppen, Villano, Doignon, & Johannesen, 1990) provides a foundation for
personalising the learning experience. The theory, in its original formalisa-
tion, is purely behaviouristic. Various approaches have been devised in order
to theoretically explain the observed behaviour by considering underlying
cognitive constructs (e.g. Falmagne et al., 1990). These approaches focus on
items’ difficulty components, their underlying demands, and skills or com-
petencies, and processes for performing them.
International Jl. on E-Learning (2006) 5(1), 75-88
IJEL 5/1 page layout 11/30/05 9:11 AM Page 75
The following section will give an introduction to the basic concepts of
Knowledge Space Theory. Subsequently, an extension of Knowledge Space
Theory is suggested as a formal framework that can serve as a basis for
implementing personalised learning into a technology-enhanced learning
system. This approach incorporates explicit reference to underlying skills
and competencies and integrates learning objects into an originally behav-
iouristic formal psychological theory with its focus on knowledge assess-
ment. Its discussion covers the derivation of skills and their structure from
ontological information, and elaborates on the impact of skill assignments
on both the assessment problems and the learning objects. It is shown that
these assignments induce structures, which allow for designing efficient pro-
cedures for adaptive assessment of the learner’s competencies, and for gen-
erating personalised learning paths.
BASIC NOTIONS OF KNOWLEDGE SPACE THEORY
Knowledge Space Theory provides a set-theoretic framework for repre-
senting the knowledge of a learner in a certain domain, which is charac-
terised by a set of assessment problems (subsequently denoted by Q). In this
framework the knowledge state of an individual is identified with the set of
problems the person is capable of solving. Due to mutual (psychological)
dependencies between the problems not all potential knowledge states (i.e.,
subsets of problems) will actually be observed. If a correct solution to a cer-
tain problem can be inferred given another problem is mastered, then each
knowledge state will contain the first problem whenever it contains the sec-
ond one (i.e. the first problem may be considered a prerequisite to the sec-
ond). To capture the relationships between the problems of a domain the
notion of a surmise relation was introduced. Two problems aand bare in a
surmise relation whenever from a correct solution to problem bthe mastery
of problem acan be surmised. A surmise relation can be illustrated by a so-
called Hasse diagram (see Figure 1 for an example), where descending
sequences of line segments indicate a surmise relation. According to the sur-
mise relation shown in Figure 1, from a correct solution to problem bthe
correct answer to problem acan be surmised, while the mastery of problem
eimplies correct answers to problems a, b, and c. A surmise relation restricts
the number of possible knowledge states and forms a quasi-order on the set
of assessment problems.
The collection of possible knowledge states of a given domain Qis called
a knowledge structure, whenever it contains the empty set Ø and the whole
set Q. The knowledge structure K induced by the surmise relation depicted
in Figure 1 is given by
K= { Ø, {a}, {c}, {a, c}, {a, b}, {a, b, c}, {a, b, d}, {a, b, c, e}, {a, b, c, d}, Q}.
76 Heller, Steiner, Hockemeyer, and Albert
IJEL 5/1 page layout 11/30/05 9:11 AM Page 76
The possible knowledge states are naturally ordered by set-inclusion,
which results in the diagram shown in Figure 2.
Figure 2 illustrates that there are various possible learning paths for mov-
ing from the naive knowledge state (empty set Ø) to the knowledge state of
full mastery (set Q). One of the possible learning paths is indicated by arrows
describing the possible steps of a learning process. It suggests to initially pre-
sent material related to problem a(or, equivalently, c), followed by material
related to problems bor c(a, respectively), and so on. Notice that the knowl-
edge structure of Figure 2 is somehow special, as it allows for gradual learn-
ing. On the one hand, each knowledge state (except state Q) has at least one
Competence-Based Knowledge Structures for Personalised Learning 77
Figure 1. Example of a Hasse diagram illustrating a surmise relation on the
knowledge domain Q= {a, b, c, d, e}
Figure 2. Knowledge structure Kinduced by the surmise relation of Figure 1.
The dashed arrows indicate a possible learning path.
IJEL 5/1 page layout 11/30/05 9:11 AM Page 77
immediate successor state that comprises all the same problems plus exactly
one. On the other hand, each knowledge state (except state Ø) has at least one
predecessor state that contains exactly the same problems, except one. A
knowledge structure with these properties, in which learning can take place
step by step, is called well-graded. According to Figure 2, for instance, the
states {a, b, c, d} and {a, b, c, e} are the immediate successor states to the
knowledge state {a, b, c}. The set {d, e} constitutes the so-called outer fringe
of the knowledge state {a, b, c}. It consists of exactly those problems that a
learner having knowledge state {a, b, c} should tackle next, and can thus
form a basis for generating personalised learning paths. The knowledge state
{a, b, c} has also two predecessor states, which are {a, b} and {a, c}. The set
{b, c} represents the so-called inner fringe of the knowledge state {a, b, c}.
Its problems may be seen as corresponding to the most sophisticated content
that has been learned recently. This is the content that the learner should
revisit, when previously learned material is to be reviewed.
Besides providing the information relevant for generating personalised
learning paths, a knowledge structure is at the core of an efficient adaptive
procedure for knowledge assessment. It allows for uniquely determining the
knowledge state by presenting the learner with only a subset of the problems
(for more details see “Problem-Based Skill Assessment”).
COMPETENCE-BASED EXTENSIONS OF KNOWLEDGE SPACE THEORY
Although there is a commercial learning system that is based on Knowl-
edge Space Theory, which is the ALEKS system (http://www.aleks.com), this
approach suffers from its limitation to a purely behaviouristic perspective. In
its original formalisation, Knowledge Space Theory focuses completely on
the observable solution behaviour, and does not refer to both learning objects
and skills or competencies that are to be taught. To overcome these limita-
tions Knowledge Space Theory may be extended so that it incorporates
explicit reference to learning objects and underlying skills and competencies.
The subsequent considerations are based on previous work by Falmagne et
al. (1990), Doignon (1994), Düntsch and Gediga (1995), Korossy (1997,
1999), Albert and Held (1994, 1999), Hockemeyer (2003), and Hockemeyer,
Conlan, Wade, and Albert (2003). It not only integrates these different con-
tributions, but also derives their implications for implementing a personalised
learning system, and clarifies the role of domain ontologies.
Extended Knowledge Space Theory is dealing with three different sorts
of entities, which are:
1. the set Qof assessment problems,
2. the set Lof learning objects (LOs),
3. the set Sof skills relevant for solving the problems, and taught by
the LOs.
78 Heller, Steiner, Hockemeyer, and Albert
IJEL 5/1 page layout 11/30/05 9:11 AM Page 78
Notice that the skills in the set Sare meant to provide a fine-grained, low-
level description of the learner’s capabilities. Usually, it is a whole bunch of
skills that is tested by an assessment problem, or taught by a LO.
Each of these basic sets is assumed to be endowed with a structure, which
we conceive as a collection of subsets of the respective set. In particular, we
consider
• a knowledge structure on the set Qof assessment problems,
• a learning structure on the set Lof LOs,
• a competence structure on the set of skills S.
As outlined, the knowledge structure constitutes the collection of possi-
ble knowledge states and forms the basis of the problem-based assessment
of a student’s competency (see “Problem-based Skill Assessment”). Usage
of the notion “competency” in the present context is in line with the termi-
nology of Doignon and Falmagne (1999), which refers to subsets of skills
that are collected in the competence structure, and which may also be called
competence states. A competence structure may either be explicitly estab-
lished by identifying prerequisite relationships between skills (see “Deriving
Skills and their Structure from Domain Ontologies”) that restrict the set of
possible competence states, or it may be indirectly induced by assigning
skills to assessment problems (or LOs) (see “Assigning Skills to Assessment
Problems” and “Assigning Skills to Learning Objects”). The learning struc-
ture together with a student’s current competence state is used to generate a
personalised learning path. Learning and competence structures are defined
in complete analogy to the knowledge structure previously introduced. Now,
the main goal is to identify the pieces of information that are needed for
establishing those structures.
SKILLS AND SKILL ASSIGNMENTS
Deriving Skills and their Structure from Domain Ontologies
This section addresses the question of how to identify skills that are rel-
evant and suitable for modelling the underlying constructs of assessment
problems and learning object regarding a certain domain. As an alternative
to cognitive task analysis (Korossy, 1999), querying experts (Zaluski, 2001),
and systematic problem construction by applying the component-attribute
approach (Albert & Held, 1994), we propose to utilise information coming
from domain ontologies.
An ontology allows structuring a domain of knowledge with respect to its
conceptual organization. It constitutes a specification of the concepts in a
domain and the relations among them and thus, defines a common vocabu-
lary of the knowledge domain. A common and natural way of representing
ontologies is by concept maps. The ontological information provided by a
Competence-Based Knowledge Structures for Personalised Learning 79
IJEL 5/1 page layout 11/30/05 9:11 AM Page 79
concept map can be used for identifying skills and for establishing a compe-
tence structure, respectively. In the sequel we outline two approaches, which
differ with respect to the level of granularity of the underlying concept map.
Identifying skills with substructures of a concept map. Skills in terms of
competence-based Knowledge Space Theory may be identified with sub-
structures of a concept map representing the ontological information of the
respective domain. This actually assumes a quite fine-grained representa-
tion, as it is necessary for a detailed characterisation of learning content, for
example. A specific skill that is required for solving problems, or that is
taught by learning objects, can be identified with a subset of the propositions
represented by the concept map. Consider, for instance, the knowledge
domain of right triangles. Figure 3 illustrates a possible assessment problem
from this domain.
Solving this geometry problem requires to know the Pythagorean Theo-
rem and how to apply it. Knowing the Pythagorean Theorem may be
assumed to constitute a skill, which corresponds to a substructure of a con-
cept map. Figure 4 provides an exemplary concept map that highlights the
substructure representing this skill. Note that not all the relevant skills can
be constructued in this way. The ability of applying the Pythagorean Theo-
rem, for example, may be regarded as a related, but separate skill, which has
to be added to the set of considered skills.
The representation of skills in the concept map may also be used for
deriving dependencies between skills, e.g. by set inclusion. If the represen-
tation of a skill xin the concept map is a subset of that of a skill y, then skill
xconstitutes a prerequisite to skill y.
Using the component-attribute approach. Concept maps provide a tool for
modelling the content of a knowledge domain, which is an essential part of
curriculum and content analysis. Within this context the construction of con-
cept maps aims at uncovering the prerequisite relations among the basic con-
cepts within a topic, and between different topics of a subject. Such a con-
80 Heller, Steiner, Hockemeyer, and Albert
Figure 3. Example of an assessment problem for the knowledge domain
“right triangles”
IJEL 5/1 page layout 11/30/05 9:11 AM Page 80
Competence-Based Knowledge Structures for Personalised Learning 81
cept map most probably will contain concepts on a higher level of abstrac-
tion, for example, Theorem of Pythagoras. This is in contrast to the more
fine-grained concept map presented before, which also captures the defini-
tion or content of these general concepts.
Curriculum and content analysis not only reveal the basic concepts of a
domain, but also the learning objectives that are related to these concepts.
Learning objectives include required activities of the learner and may be
captured by so-called action verbs. Action verbs (e.g., state, or apply a the-
orem) describe the observable student performance or behaviour and may be
annotated to the nodes of the concept map representing the concepts that are
to be taught. The information provided by the concept map then again can
be used for establishing a competence structure in the sense of Knowledge
Space Theory.
The concept map provides a hierarchical structure on the concepts of a
domain. For instance, according to the curriculum the Pythagorean Theorem
constitutes a prerequisite to the Altitude Theorem. This induces an order on
the set of concepts C. The relation between the concepts may be represented
graphically as in Figure 5(a). Additionally, a relation may be introduced on
the set of action verbs Athat induces a structure on it. For instance, to “state”
a particular theorem is most likely a prerequisite to “apply” the respective
theorem, and therefore, the action verb “state” can be considered as a pre-
requisite to the action verb “apply.” The structure defined on the action verbs
can also be illustrated by a graph (see Figure 5(b) for an example).
Figure 4. Concept map of the knowledge domain “right triangles.” The
marked substructure refers to the skill “knowing the Pythagorean
Theorem.”
is half
the product of
cuts into
2 right triangles
height h is perpendicular
to
is opposite of
right angle
hypotenuse c
squared
has attribute
has attribute right
triangle
has leg has leg
longest leg of cathetus a
square of
hypotenuse c2
squared
square of
cathetus a2
has leg
has
attribute
is adjacent to
area A
is half
the product of cathetus b
is adjacent to
squared
square of
cathetus b2
equals
sum of
IJEL 5/1 page layout 11/30/05 9:11 AM Page 81
Based on these considerations, a skill in terms of extended Knowledge
Space Theory may be identified with a pair consisting of a concept and an
action verb (e.g. c1a2). As an example for a skill consider “apply the Pythagore-
an Theorem,” which consists of the concept “Pythagorean Theorem” and the
action verb “apply.” Formally we define the set of skills by S⊆C ×Ato reflect
the fact that not all combinations of concepts and action verbs may be mean-
ingful, or even realisable. A crucial question is how to merge the two kinds of
structures, that is, the structure on the set of concepts and the structure on the
set of action verbs, to establish a structure on the set of skills.
To resolve this issue we suggest the component-attribute approach
(Albert & Held, 1994, 1999). According to this approach components are
understood as dimensions, while attributes are the different values these
dimensions can take on. In the present context, the set Cof concepts and the
set Aof action verbs are considered as the components, and the attributes are
identified with the respective elements (e.g. c1, c2, c3, c4in Cand a1, a2in A).
On each component a relation is defined that orders the attributes (see Fig-
ure 5). A structure on the set of skills is then established by forming the
direct product of these two components, which results in a prerequisite rela-
tion on the Cartesian product C × A. The product of the two graphs displayed
in Figure 5 is the relation depicted in Figure 6. From this you can see, e.g.
that skill c2a2is a prerequisite to the skills c2a1, c1a1, and c1a2, but to none of
the other skills.
If Sis a proper subset of the Cartesian product C×Athen we consider the
prerequisite relation that the direct product shown in Figure 6 induces on S.
In the framework of extended Knowledge Space Theory the prerequisite rela-
tion on the skills is interpreted as a surmise relation that gives rise to the com-
petence structure. The competence states contained in it have to respect the
ordering illustrated in Figure 6, which means, for example, that with the skill
c3a1each competence state has to contain the skills c3a2, c4a1, and c4a2, too.
82 Heller, Steiner, Hockemeyer, and Albert
Figure 5. Concept structure (a) and structure defined on action verbs (b)
IJEL 5/1 page layout 11/30/05 9:11 AM Page 82
Note, that from a psychological point of view, pairs consisting of a con-
cept and an action verb, like “state Pythagorean Theorem” or “apply
Pythagorean Theorem,” describe rather global skills. Applying the
Pythagorean Theorem might require several more elementary skills, which
are in correspondence with the distinct steps in a solution path (e.g., extract-
ing a root, transforming). It may thus be necessary to characterise the skills
at a more fine-grained level. Further research is needed to decide upon an
optimal level of granularity of the skills.
Assigning Skills to Assessment Problems
Let us now consider the assignment of skills to the set of assessment
problems. The relationship between assessment problems and skills can be
formalised by two mappings.
• The mapping s(skill function) associates to each problem a collection
of subsets of skills. Each of these subsets (i.e., each competency) con-
sists of those skills that are sufficient for solving the problem. Assign-
ing more than one competency to a problem takes care of the fact that
there may be more than one way to solve it.
• The mapping p(problem function) associates to each subset of skills the
set of problems that can be solved in it. It defines a knowledge structure
because the associated subsets actually are nothing else but the possible
knowledge states.
It has been shown that both notions are equivalent (Düntsch & Gediga,
1995), which means that, given the skill function, the problem function is
uniquely determined, and vice versa. Consequently, only one of the two
functions needs to be known to build the respective knowledge structure.
Consideration is confined to the skill function, because it may be interpret-
ed as representing the assignment of metadata to the problems. It follows
Competence-Based Knowledge Structures for Personalised Learning 83
Figure 6. Example of a prerequisite relation on the skills induced by the
structures on concepts and action verbs displayed in Figure 5.
IJEL 5/1 page layout 11/30/05 9:11 AM Page 83
that assigning (semantic) metadata to assessment problems puts constraints
on the possible knowledge states that can occur.
We illustrate the intimate relationship between skill function and problem
function by a simple example. Consider the knowledge domain Q= {a, b, c, d},
and let the skill function son the set S= {x, y, z} of skills be given by
s(a) = {{x, y},{x, z}}, s(b) = {{x, z}}, s(c)= {{x},{y}}, s(d) = {{y, z}}.
This means, for example, that each of the skill sets {x, y} and {x, z} is
sufficient for solving problem a. From the skill function we can derive the
corresponding problem function, which yields
p(∅) = ∅,p({x}) = {c}, p({y}) = {c}, p({z}) = ∅,
p({x,y}) = {a, c}, p({x,z}) = {a, b, c}, p({y,z}) = {c, d}, p(S) = Q.
The assignment of skills to the assessment problems induces a knowledge
structure on the set of problems, which is actually given by the subsets of
problems in the range of the problem function. The knowledge structure for
these examples is given by {∅, {c}, {a, c}, {c, d}, {a, b, c}, Q}. Whenever
a competence structure is available, e.g. as a result of exploiting ontological
information (see “Deriving Skills and their Structure from Domain Ontolo-
gies”), the domain of the problem function is restricted to the actually occur-
ring competence states. This puts additional constraints on the set of possi-
ble knowledge states.
In principle, the skill function for a given set Qof assessment problems
may introduce dependencies between skills, too. It may be the case that a cer-
tain skill is required for solving a problem only in connection with another
skill. In the above example the skill zis available only if either xor yis avail-
able. These dependencies, however, may only crop up in the given set Q, and
it remains unclear whether they are valid in general. If capitalising on inci-
dental dependencies between problems is to be avoided then the constraints
the skill function puts on the possible subsets of skills should be neglected.
Problem-Based Skill Assessment
A knowledge structure can form the basis for devising an efficient adap-
tive procedure for knowledge assessment (Doignon & Falmagne, 1999;
Dowling & Hockemeyer, 2001). Problem-based skill assessment proceeds in
two steps. First, the knowledge state of a learner, which refers to the observ-
able behaviour, is adaptively assessed. After identifying a learner’s knowl-
edge state, the knowledge state can be mapped to the corresponding compe-
tence state in a second step.
Considering the knowledge structure given in Figure 2 for the knowledge
domain Q= {a, b, c, d, e}, in the beginning of an assessment phase all states
of the structure may correspond to the knowledge state of an individual learn-
84 Heller, Steiner, Hockemeyer, and Albert
IJEL 5/1 page layout 11/30/05 9:11 AM Page 84
er. According to a deterministic procedure, the assessment starts by selecting
a problem that is contained approximately in half of the states of this struc-
ture and by posing this problem to the learner. Dependent on the learner’s
answer, the next problem will be selected. If the learner is capable of solving
problem b, for example, then only the knowledge states containing problem
bare still feasible. If subsequently problem eis solved, states {a, b, c, e} and
{a, b, c, d, e} remain. The learner’s knowledge state is uniquely identified
after presenting problem d. For instance, state {a, b, c, e} results if problem
dcannot be solved by this learner. Thus, for a set of five assessment prob-
lems, the presentation of only three problems allows for identifying the
knowledge state of a learner. Formally, the number of questions for deter-
mining the knowledge state of a learner is approximately the dual logarithm
of the total number of knowledge states.
Aside from the outlined deterministic assessment procedure, assessment
may also be embedded into a probabilistic framework. A probabilistic
assessment method allows for considering that the knowledge states may
occur with different frequencies within a population as well as that a subject
sometimes may be careless in answering a problem or may guess the correct
answer. Such an assessment method assumes an a priori likelihood function
(e.g. probability distribution) on the knowledge states. Initially, this likeli-
hood may depend on the learner’s profile, for example, the age, or grade of
this learner. Later, this probability distribution is updated consistent with the
learner’s answers to the posed problems. The questioning continues until
there is a pronounced peak in the likelihood function that suggests a unique
knowledge state for an individual learner.
The knowledge state identified for a learner then can be mapped to his/her
competence state by using the skill function. This means that, given a knowl-
edge state, we are looking for the subset of skills that are sufficient for solving
the problems contained in the knowledge state. However, there may be more
than one such subset. In this case the skills cannot be recovered uniquely given
the assessed knowledge state. To provide an example, consider the skill func-
tion defined in “Assigning Skills to Assessment Problems.” If we assume that
the assessment converged to the knowledge state {c} then it is unclear, which
skills the learner is endowed with. According to the skill function either skill x
or skill ymay be responsible for solving problem c. This nonuniqueness occurs
whenever a problem function is not one-to-one. Using additional information
may lead to a unique identification of the available skills (e.g. looking up the
learning history, checking for the skills actually taught). The best strategy, how-
ever, would be to select a proper set of assessment problems that avoids the
nonuniqueness. Once the competence state of a learner has been determined it
may serve as a basis for selecting a personalised learning path.
Competence-Based Knowledge Structures for Personalised Learning 85
IJEL 5/1 page layout 11/30/05 9:11 AM Page 85
Assigning Skills to Learning Objects
The relationship between learning objects and skills is different from that
between assessment problems and skills. The relationship between the set L
of LOs and the skills in Sis mediated by two mappings (Hockemeyer, 2003;
Hockemeyer et al. 2003). The mapping r associates to each LO a subset of
skills (required skills), which characterise the prerequisites for dealing with
it, or understanding it. The mapping tassociates to each LO a subset of skills
(taught skills), which refer to the content actually taught by the LO. In a sim-
ilar way as previously outlined, the mappings rand t induce a learning struc-
ture on the set of LOs, which plays a central role for generating personalised
learning paths. The pair of mappings rand talso imposes constraints on the
competence states that can occur. Again, these constraints are tied to the
given set Lof LOs. The imposed competence structure characterises the
learning progress that may be achieved by studying the learning objects in L.
Generally, the assignment of skills to learning objects allows for deciding
upon which learning objects are to be presented next, given a certain com-
petence state. The concepts of inner and outer fringes (see “Basic Notions of
Knowledge Space Theory”) of a competence state may provide the basis for
implementing personalised learning. The inner fringe of a competence state
may be interpreted as “what a learner can do,” while the outer fringe repre-
sents “what this learner is ready to learn.” Therefore, proceeding in the
learning process the next skills to be learned should be chosen from the outer
fringe of the current competence state. Thus, a suitable learning object has
to be selected that is characterized by required skills that the learner has
already available and by taught skills that correspond to the outer fringe of
the current competence state. If previously learned material has to be
reviewed, then the content corresponding to the inner fringe of a learner’s
actual competence state seems to be a natural choice, because it contains the
most sophisticated skills acquired by the learner.
CONCLUSIONS
The present article proposes a competence-based extension of Knowl-
edge Space Theory that provides a formal framework for explicitly linking
assessment problems and learning objects to the relevant skills and compe-
tencies. It is demonstrated that the assignment of skills to assessment prob-
lems (which are sufficient for their solution) induces a knowledge structure
characterising the possible answer patterns of the learners. Moreover, it is
shown that assigning required and taught skills to learning objects allows for
generating personalised learning paths. Introducing skills provides a gener-
al framework for relating models of the domain, the learners, and the learn-
ing objects, as described by Bouzeghoub, Defude, Duitama, and Lecocq
(2006, this issue). These authors also refer to information about what is
86 Heller, Steiner, Hockemeyer, and Albert
IJEL 5/1 page layout 11/30/05 9:11 AM Page 86
required and what is provided by a LO, which is perfectly in line with the
assignment of required and taught skills to LOs as discussed in “Assigning
Skills to Learning Objects.” The proposed skill assignments also contribute
to the reusability of LOs (see Strijker & Collis, 2006, this issue).
The article provides a detailed discussion of how to derive relevant skills
and their structure from domain ontologies. Two possible approaches are out-
lined. On the one hand, skills are identified with substructures of a concept
map. On the other hand, skills are identified with pairs of concepts and action
verbs, and a skill structure is established by merging the structures given on
both sets. Assigning these skills to assessment problems and LOs, as sug-
gested by the competence-based extension of Knowledge Space Theory,
yields a framework for an efficient adaptive assessment of the skills and com-
petencies of a learner, and for selecting personalised learning paths. This
framework constitutes a valuable model for implementing personalised learn-
ing within an open technology-enhanced learning system. The implementa-
tion of the outlined theoretical framework within the iClass project is dis-
cussed by Türker, Görgün, and Conlan (2006, this issue), while Brady, O'Ke-
effe, Conlan, and Wade (2006, this issue) focus on the personalisation of the
presented learning material via skill- or concept-based services offered by the
Selector and the LO Generator module of the iClass system. A discussion of
how to handle and integrate multiple skill assignments that characterize (par-
tially overlapping) learning material coming from distributed resources is
contained in Heller, Mayer, Hockemeyer, and Albert (2005).
References
Albert, D., & Held,T.(1994). Establishing knowledge spaces by systematical problem construction.
In D. Albert (Ed.), Knowledge structures (pp. 78–112). New York: Springer Verlag.
Albert, D., & Held,T.(1999). Component based knowledge spaces in problem solving and inductive
reasoning. In D. Albert & J. Lukas (Eds.), Knowledge spaces: Theories, empirical research
applications (pp. 15–40). Mahwah, NJ: Lawrence Erlbaum.
Bouzeghoub, A., Defude, B., Duitama, J.F., & Lecocq, C. (2006).A knowledge-based approach
to describe and adapt learning objects. International Journal on E-Learning, 5(1).
Doignon, J. (1994).Knowledge spaces and skill assignments. In G. H.Fischer & D. Laming (Eds.),
Contributions to mathematical psychology, psychometrics and methodology (pp. 111–121).
New York: Springer Verlag.
Doignon, J.P., & Falmagne, J.C. (1985). Spaces for the assessment of knowledge. International
Journal of Man-Machine Studies, 23, 175-196.
Doignon, J.P. & Falmagne, J.C. (1999). Knowledge spaces. Berlin: Springer.
Dowling, C. E., & Hockemeyer, C. (2001). Automata for the assessment of knowledge. IEEE
Transactions on Knowledge and Data Engineering, 13, 451–461.
Düntsch, I., & Gediga, G. (1995). Skills and knowledge structures. British Journal of Mathematical
and Statistical Psychology, 48, 9–27.
Competence-Based Knowledge Structures for Personalised Learning 87
IJEL 5/1 page layout 11/30/05 9:11 AM Page 87
Falmange, J.C., Koppen, M., Villano, M., Doignon, J.P. & Johannesen, L. (1990). Introduction to
knowledge spaces: How to build, test,and search them. Psychological Review,97, 201-224.
Heller, J., Mayer, B., Hockemeyer, C., & Albert, D. (2005). Competence-based knowledge
structures for personalised learning: Distributed resources and virtual experiments. In G.Albano, P.
Ritrovato, & S. Salerno (Eds.), Proceedings of the 1st International EleGI Conference on Advanced
Technology for Enhanced Learning. Swindon Wiltshire, UK: The British Computer Society (BCS).
Hockemeyer, C. (2003). Competence based adaptive e-learning in dynamic domains. In F. W.
Hesse & Y.Tamura (Ed.), The Joint Workshop of Cognition and Learning through Media-Com-
munication for Advanced E-Learning (JWCL), 2003, 79–82, Berlin.
Hockemeyer, C., Conlan, O., Wade, V., & Albert, D. (2003). Applying competence prerequisite
structures for eLearning and skill management. Journal of Universal Computer Science, 9,
1428-1436.
Korossy, K. (1997). Extending the theory of knowledge spaces: A competence-performance
approach. Zeitschrift für Psychologie, 205, 53-82.
Korossy, K. (1999). Modeling knowledge as competence and performance. In D. Albert & J.
Lukas (Eds.), Knowledge spaces: Theories, empirical research, applications (pp. 103-132).
Mahwah, NJ: Lawrence Erlbaum.
O’Keeffe, I., Brady, A., Conlan, O., & Wade, V. (2006). Just-in-time generation of pedagogically
sound, context sensitive personalized learning experiences. International Journal on E-Learn-
ing, 5(1).
Strijker, A., & Collis, B. (2006). Strategies for reuse of learning objects: Context dimensions.
International Journal on E-Learning, 5(1).
Türker, A., Görgün, I., & Conlan, O. (2006). The challenge of content creation to facilitate
personalized eLearning experiences. International Journal on E-Learning, 5(1).
Zaluski, A.(2001). Knowledge spaces mathematica package. In Proceedings of PrimMath 2001
- Mathematica in Science, Technology and Education. Zagreb,Croatia, University of Zagreb.
Acknowledgement
The work presented in this paper is partially supported by the European
Community under the Information Society Technologies (IST) program of
the 6th FP for RTD – project iClass contract IST-507922. The authors are
solely responsible for the content of this paper. It does not represent the
opinion of the European Community, and the European Community is not
responsible for any use that might be made of data appearing therein.
88 Heller, Steiner, Hockemeyer, and Albert
IJEL 5/1 page layout 11/30/05 9:11 AM Page 88
