Using Visual Guidance and Feedback Based on
Competence Structures for Personalising
Dietrich Alberta, Alexander Nussbaumera, Christina Steinera
aDepartment of Psychology, University of Graz, Austria
Abstract: In this paper a novel approach is presented how visualisation of knowledge and
competence structures can be employed in order to support personalisation in e-learning
systems. Information visualisation strategies are used to present ontology-based domain
knowledge and related competence structures to the learner. In contrast to adaptive
e-learning systems which employ domain and user models for the adaptation and
personalisation strategies, the learner uses these visual knowledge maps to decide on the
own learning process. Furthermore, visual knowledge maps are used to give feedback to the
learner about learning progress and assessment result. Methods and techniques are presented
how these maps can be used for visual guidance and feedback. A set of learning tools have
been developed which implement and demonstrate the proposed approach.
Keywords: Adaptive system, adaptivity, self-regulated learning, skill, competence,
Knowledge Space Theory, information visualisation, open learner model
Adaptivity and personalisation provided by adaptive systems capable of tailoring content
and behaviour to characteristics and needs of learners has been an important research strand
in e-learning for a long time. This research field has its origin in technological developments
(computer systems, Internet, hypermedia) and is characterised by research how technology
can support and guide the learning process. However, this approach to e-learning holds the
risk of having the learning process to a large extent controlled by the system. Furthermore, if
the models or structures underlying system behaviour are invalid, then guidance provided
by the system is actually worse than no guidance (De Bra, 2000).
The approach presented in this paper makes use of competence-based models and
structures in a different way than adaptive systems employ them for determining system
behaviour. In contrast to traditional adaptive systems, these models are not hidden from the
user and only used by the adaptation algorithms, but - and this is seen as the major
innovation of this paper - these models are presented to the learner directly in a visual and
interactive way. In contrast to some experiences with open learner models in the last years
(Bull & Kay, 2008), our approach focuses on visualising skill-based models. With the help
of these models the learner gets both more control over the own learning process and
specific information about the own learning progress. Instead of traditional system guidance
the learner gets visual guidance and feedback which does not limit the learner to automated
decisions of the adaptive systems.
A set of learning tools has been developed which follow and demonstrate this
approach. Research and development of these tools are part of the iClass research project
(iClass, 2008). The aim was to support a self-regulative learning cycle, which according to
(Zimmerman, 2002) consists of forethought (planning), performance (monitoring) and
reflection. The developed tools support the planning and reflection processes, performance
(viewing learning objects) is done by the use of other iClass components.
The next section gives an overview on the research fields which are basis for our
approach. Section 2 gives a more detailed description of the knowledge representation
model which is used for our approach. Section 3 explains principles how guidance and
feedback can be realised in a visual way and Section 4 presents some developed learning
tools which implement those principles. Future work and conclusion can be found in
1. Theoretical Foundation and Background
1.1 Competence-based Knowledge Space Theory
Knowledge Space Theory (KST) and its competence-based extensions (CbKST) are
prominent examples how an adaptation strategy can be grounded on a theoretical
framework (Hockemeyer, 2003). KST constitutes a sound psychological mathematical
framework for both structuring knowledge domains and for representing the knowledge of
learners (Albert et al., 1999). Due to (psychological) dependencies between problems
prerequisite relations can be established. The knowledge state of a learner is identified with
the subset of all problems this learner is capable of solving. By associating assessment
problems with learning objects, a structure on learning objects can be established, which
constitutes the basis for meaningful learning paths adapted to the learners knowledge state.
Competence-based Knowledge Space Theory (CbKST) incorporates psychological
assumptions on underlying skills and competencies that are required for solving the
problems under consideration (Korossy, 1996 and Heller et al. 2006). This approach assigns
to each problem a collection of skills which are needed to solve this problem and to each
learning objects those skills which are taught. Similar to the knowledge state a competence
state can be defined which consists of a set of skills which the learner has available.
Furthermore, there may also be prerequisite relationships between skills.
CbKST provides algorithms for efficient adaptive assessment to determine the
learner's current knowledge and competence state, which builds the basis for
personalization purposes. Based on this learner information, personalised learning paths can
be created. Goal setting can be done by defining skills to be achieved (competence goal) or
problems to be capable of solving. The competence gap to be closed during learning is
represented by the skills which are part of the goal but not part of the competence state of a
1.2 Adaptive Systems and Guidance
The concept of adaptivity has a long tradition in technology-enhanced learning, for example
it has been applied in Intelligent Tutoring Systems (ITS), user-model-based Adaptive
Systems (AS), and Adaptive Hypermedia Systems (AHS) (Brusilovsky, 2000). Following
the discussion in (Brusilovsky, 1996 and De Bra et al., 2004), users (learners) differ in terms
of (learning) goals, pre-knowledge, individual traits and needs, as well as pedagogical
parameters. Based on these characteristics adaptive presentation (adaptation on the content
level) and adaptive navigation support (direct guidance, adaptive ordering, hiding, and
annotation of links) are the most important features which can be provided by an adaptive
system. Domain models and user models are defined in order to specify relationships
between users and content, which forms the basis for the adaptation functionality. In
educational applications these relationships typically represent the knowledge about
learners and content. Furthermore, adaptive systems usually contain adaptation models
which determine the adaptation strategy of those systems. In this way an adaptive system
can help the learner to navigate through a course by providing user-specific paths.
1.3 Information and Knowledge Visualisation
The abilities of humans to recognise visual information are highly developed. Patterns,
colours, shapes and textures can rapidly and without any difficulty be detected. On the other
hand, the perception of text-based content is much more effort than the perception of visual
information (Shneiderman, 1996). Information visualisation is the transformation of
abstract data and information into a form that can be recognised and understood by humans.
In this sense, information visualisation can be seen as an interface to abstract information
spaces. So exploring large volumes of data can be done effectively by humans.
Information visualisation techniques are widely used in Web-based social software
(e.g. graph visualisation is used to outline online community networks and tag clouds are
often used to provide overview on collaboratively tagged Web content) and especially in
knowledge management (e.g. visualisation of large knowledge structures for providing
overview and interface to it). In (Herman et. al, 2000) an overview and survey on graph
visualisation techniques is provided. In contrast to these application areas, information
visualisation is barely used in e-learning applications. For example, in (Kay et al., 2007)
visualisation techniques are described how visual feedback and mirroring can be provided
to group and collaborative learning.
2. Knowledge Representation Model
The knowledge representation model elaborated in the context of the iClass project is based
on two pillars: The first pillar is the domain model which represents the knowledge of the
domain and the second pillar is the user model which represents the knowledge of the
learner. The knowledge representation model for a single domain is depicted in Figure 1.
The domain model is based on CbKST and comprises the structure how the domain is
represented regarding learning objects, problems (assessment items), skills, concepts, and
action verbs. The central elements are skills which are assigned to both learning objects and
assessment items. A skill is defined by a set of domain concepts and an action verb which
specifies the level of expertise regarding the respective concepts (e.g. apply the Pythagorean
Theorem). The prerequisite relations between skills build the competence structure. The
domain model is based on the model described in (Görgün et al., 2005), and has been
slightly simplified in order to allow teachers to easily create their own models of the
domains they teach.
A domain map represents a knowledge domain and consists of domain information
populated in line with the structure of the domain model. This can be done by content
authors as well as by teachers. Domain maps can be differentiated between subjective and
objective maps to indicate their state of validation. Objective maps are empirically validated
(ideal case) and subjective maps are not validated, even if they have been created by domain
The learner model characterises knowledge, competences and learning goal of a
learner. It refers to elements of the domain model, since the learner's knowledge is defined
through references to domain knowledge. The knowledge state is a set of problems which
the learner is capable of solving, the competence state is a set of skills which the learner has
available, and the competence goal is a set of skills which the leaner should acquire.
Learner Model Domain Model
Learner Knowledge Representation Domain Knowledge Representation
Figure 1: Knowledge Representation Model consisting of domain model and learner model.
3. Visual Guidance and Feedback
The basic idea how to provide learner and teacher support through visual guidance and
feedback is to visualise the elements and relations of the knowledge representation model
which are typically used by adaptive systems for the personalisation process. An interactive
learning tool which renders the models enables the learner to visually interact with
visualised information for several purposes. For example, plans and sequences of learning
objects and their related skills can be created, learning progress and assessment result can be
reported, and a history of the learning experience can be displayed. The learner gets an
overview on both the domain and the own learning progress.
For example, a visual representation of the structure (prerequisite relation) on skills
can be used for both guidance and feedback. In Figure 2 such a structure is depicted as
Hasse diagram for six skills, with ascending sequences of line segments representing
prerequisite relationships. On this structure a learner can see what should be learned, i.e. the
six skills making up the competence goal. Due to the prerequisite relation, it also can be
seen, that there is an order how the skills should be acquired. Before addressing
higher-ordered skills, the relevant prerequisite skills should be learned. Since skills are
assigned to learning objects, learning objects can be searched which teach the respective
skills. In this way, a plan consisting of a meaningful sequence of learning objects can be
created. After the learning process, an assessment can be conducted which detects the
acquired skills of a learner and the result can again be displayed in the visualisation of the
competence structure. In this way, the learner can e.g. reflect on his current competence in
relation to the competence goal (see Figure 2). To complete the learning cycle the missing
(not acquired) skills can immediately be recognised and a new plan of learning objects can
be composed which teach those skills.
As outlined in this example, using these structures the number of reasonable choices
can be significantly reduced. Information on the learner’s current competence state can be
utilised for recommending learning options from which to choose that correspond to
possible next steps of learning. In a similar way, the other elements of the knowledge
representation model can be utilised. Concepts and action verbs can visually be related to
skills which supports the understanding of skills.
Figure 2: Structure on skills illustrated as Hasse diagram. In the right figure
available skills of a learner are displayed.
4. Implemented Learning Tools
The principles of visual guidance and feedback described above have been implemented in
the iClass system in various tools to support the learning and teaching process.
4.1 Skill-based Planning Tool
The Planning Tool (Figure 3) supports the learning processes of goal setting and use of task
strategies. This tool visualises the domain skills and their prerequisite relation as Hasse
diagram. On this graph, skills can be chosen to define the competence goal and subsequently
sequenced on the visual plan component. Prerequisite skills of the chosen skills are also
added to the plan. If the created sequence of the skills is not in line with the prerequisite
structure, this tool gives visual feedback about this situation (in terms of coloured skills). It
also provides the functionality of automatically sequencing the chosen skills corresponding
to the prerequisite relations. Furthermore, for each skill learning objects can be searched and
chosen which teach the respective skill. As soon as for all skills of the competence goal
learning objects have been added to the plan, visual feedback is provided that the plan is
complete. Further guidance is granted, as the tool also can propose meaningful sequences of
learning objects by using the relations between learning object and skill.
4.2 Adaptive Assessment Tool
An adaptive assessment based on KST (Doignon and Falmagne, 1999) is conducted to
determine which skills a learner has available. Questions are posed to the learner taking into
account previous answers and exploiting prerequisite relationships among problems. The
traditional algorithm calculates the sequence of questions and is capable of posing a
minimal number of questions to determine the learner's knowledge. The result of the
assessment is a (verified) set of skills (competence state) which the learner has available,
which is subsequently visualised and reported to the learner.
4.3 Self-Evaluation Tool
A learner may reflect on what has been learned by defining skills which consist of concepts
and action verbs. This is done in three steps: (1) The learner is provided with a list of
concepts and chooses those concepts that has been covered in the learning process so far. (2)
Then for each concept the level of ‘expertise’ is to be indicated which have been acquired so
far according to the learner's estimation – these levels are indicated by the Bloom taxonomy
levels, i.e. the action verbs remember, understand, and apply. (3) The combination of
concepts and Bloom level action verbs results in skills – defined by the learner.
Figure 3: Skill-based Planning Tool visualises the competence structure where the learner can interactively
create the own learning plan.
4.4 Knowledge Representation Tool
This tool presents the skills learned during the learning process. Three sources for this
information are used: (1) The skills which have been taught by learning objects are
visualised in a chronological order together with the learning objects, (2) the acquired
(verified) skills resulted from the adaptive assessment and (3) the skills (non-verified)
resulted from the Self-evaluation Tool. The presentation of skills is done in a visual way,
learned and acquired skills (verified and non-verified) are rendered in different colours.
Furthermore, the competence goal (also set of skills) is rendered in a way that missing skills
can be seen immediately. In this way the learner directly monitors his learning progress and
skill gap (compared to the competence goal).
4.5 Domain Map Editor
All presented tools rely on the knowledge representation model consisting of skills, learning
objects, assessment question, concepts, and action verbs. Creating the domain map for a
certain knowledge domain is usually the task of teachers and domain authors. A tool has
been developed which allows for easily creating domain maps by again employing
visualisation techniques. For example, defined prerequisite relations between skills can be
immediately visualised, and the assignment of skills to learning objects is depicted in a
fish-eye visualisation where all learning objects including the assigned skills are shown and
the selected learning object is magnified (Figure 4).
Figure 4: Domain Map Editor. All leaning objects (activities) of a domain are displayed together with the
assigned skills. A fish-eye distortion magnifies the selected learning object.
5. Conclusion and Outlook
In this paper a novel approach has been presented how learning can be supported and
stimulated by using visualisation techniques. This approach makes use of concepts of
adaptive systems and related research in order to integrate guidance in learning processes. A
knowledge representation model (domain and user model) is used as a basis for the
guidance. In contrast to adaptive systems, these models are not hidden from the user and
only used by the adaptation algorithms, but - and this is seen as the major innovation of this
paper - these models are visualised by the learning tools. Through these visualisations the
learner can get both guidance and responsibility for his learning process at the same time.
Several tools have been developed which exploit this novel approach in order to support
particular self-regulated learning processes.
The presented approach is supposed to have great potential for further work and can
enrich the lively research field of information visualisation by new possibilities of visual
guidance and feedback. More relations between elements of the knowledge representation
model can be visualised and used for the further learning tools. Furthermore, the presented
approach is also supposed to have great influence to the field of self-regulated learning.
Visual guidance can stimulate the self-regulated planning process of a learner, and visual
feedback can support self-monitoring and reflection on learning
The work presented in this paper is supported by 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.
 Albert, D., & Lukas, J. (Eds.) (1999). Knowledge Spaces: Theories, Empirical Research, and
Applications. Mahwah, NJ: Lawrence Erlbaum Associates.
 Bull, S., & Kay, J. (2008). Metacognition and Open Learner Models. In: The 3rd Workshop on
Meta-Cognition and Self-Regulated Learning in Educational Technologies, at ITS2008.
 Brusilovsky, P. (1996). Adaptive Hypermedia, an Attempt to Analyze and Generalize, In P. Brusilovsky,
P. Kommers, & N. Streitz (Eds.), Multimedia, Hypermedia, and Virtual Reality (Lecture Notes in
Computer Science, Vol. 1077). Berlin: Springer-Verlag,
 Brusilovsky, P. (2000). Adaptive hypermedia: From intelligent tutoring systems to Web-based education
(Invited talk), In: G. Gauthier, C. Frasson and K. VanLehn (eds.) Intelligent Tutoring Systems. Lecture
Notes in Computer Science, Vol. 1839, pp. 1-7, Berlin: Springer Verlag.
 De Bra, P. (2000). Pros and Cons of Adaptive Hypermedia in Web-Based Education, In: Journal on
CyberPsychology and Behavior, 3(1), (2000), pp. 71-77, Mary Ann Lievert Inc.
 De Bra, P., Aroyo, L., Cristea, A. (2004). Adaptive Web-based Educational Hypermedia, In: Mark
Levene, Alexandra Poulovassilis (Eds.) Web Dynamics, Adaptive to Change in Content, Size, Topology
and Use, pp. 387-410, Springer-Verlag.
 Doignon, J.-P., & Falmagne, J-C (1999). Knowledge Spaces. Heidelberg, Berlin, New York:
 Görgün, I., Türker, A., Ozan, Y., & Heller, J. (2005). Learner Modeling to Facilitate Personalized
E-Learning Experience. Proceedings of Cognition and Exploratory Learning in Digital Age (CELDA
2005) (pp. 231-237).
 Heller, J., Steiner, C., Hockemeyer, C., & Albert, D. (2006). Competence-Based Knowledge Structures
for Personalised Learning, In: International Journal on E-Learning, 5(1), (2006), pp. 75-88.
 Herman, I., Melancon, G., & Marshall, M. (2000). Graph Visualisation and Navigation in Information
Visualisation: A Survey. IEEE Transactions on Visualization and Computer Graphics, 6(1), pp. 24-43).
 Hockemeyer, C. (2003). Competence Based Adaptive E-Learning in Dynamic Domains, In: F. W. Hesse
& Y. Tamura (Eds.), The Joint Workshop of Cognition and Learning through Media-Communication for
Advanced E-Learning (JWCL 2003) , pp. 79-82, Berlin.
 iClass (2008): The iClass research project, http://www.iclass.info/
 Kay, J., Kalina, Y., & Reinmann,.P. (2007). Visualisations for Team Learning: Small Teams Working on
Long-term Projects. Proceedings of the Conference on Computer-supported Collaborative Learning
(CSCL 2007), International Society of the Learning Sciences.
 Korossy, K. (1997). Extending the Theory of Knowledge Spaces: A Competence-Performance
Approach. Zeitschrift für Psychologie, 205, pp. 53-82.
 Shneiderman, B. (1996). The Eyes Have It: A Task by Data Type Taxonomy for Information
Visualizations, In: Proceedings of the IEEE Symposium on Visual Languages (VL 1996), pp. 336–343.
 [Zimmerman, 2002] Zimmerman, B.J.: Becoming a Self-regulated Learner: An Overview, In: Theory
into Practice, 41(2), (2002), Springer-Verlag.