Adaptive Learning Objects Sequencing for Competence-Based Learning. Pythagoras Karampiperis, Demetrios Sampson. Department of Technology. Education ...
Adaptive Learning Objects Sequencing for Competence-Based Learning Pythagoras Karampiperis, Demetrios Sampson Advanced e-Services for the Knowledge Department of Technology Society Research Unit, Education and Digital Systems, & Informatics and Telematics Institute, Centre for University of Piraeus, Research and Technology Hellas, Greece Greece e-mail:{pythk, sampson}@iti.gr
Abstract Lifelong learning refers to the activities people perform throughout their life to improve their competence in a particular field. Although adaptive educational hypermedia systems (AEHS) bare the potential to provide personalized learning experiences based on individual’s knowledge, skills, and competencies, so far they have been confined in the provision of knowledge resources that are adaptively selected based on the structure of concepts a domain model. In this paper, we address the problem of designing adaptive educational hypermedia systems for supporting lifelong competence development. To this end, we propose an alternative to the traditional subject domain specific design approach which is based on the use of competence description ontology and learner’s competence records.
individual’s knowledge, skills, and competencies, they are limited in the provision of knowledge resources that are adaptively selected based on the structure of concepts of the subject domain model (Concept Selection Rules) [5]. In this paper, we address the problem of designing adaptive educational hypermedia systems for supporting lifelong competence development. To this end, we propose an alternative to the traditional subject domain specific design approach which is based on the use of competence description ontology and learner’s competence records. The paper is structured as follows: First, we present our proposed approach for adaptive educational hypermedia systems. Finally, we present simulation results of the proposed approach and discuss our findings and the conclusions that can be offered.
2. The proposed adaptive educational hypermedia architecture 1. Introduction It is commonly recognized that lifelong learning is not just one aspect of education and training, but is a guiding principle for provision and participation across the full continuum of learning contexts [1]. Lifelong learning refers to the activities people perform throughout their life to improve their competence in a particular field [2]. Competence is defined as the integrated application of knowledge, skills, values, experience, contacts, external knowledge resources and tools to solve a problem, to perform an activity, or to handle a situation [3]. Adaptive Educational Hypermedia Systems (AEHS) attempt to personalize the learning experience for a given learner [4]. Although AEHS have the potential to provide personalized learning experiences based on
In our previous work [6], we presented a generalized architecture of an AEHS. In order to support the generation of personalized learning experiences for lifelong competence development, several components of the generalized architecture need to be appropriately modified. In the proposed architecture (presented in Fig. 1) the Subject Domain Model has been reformed to contain a hierarchy of competencies that are needed to grand a certain competence level (that is, the Competence Development Programme), as well as, an ontology (that is, the Competence Ontology) for describing a given set of competencies. For each competence level specified in the Competence Development Programme, a set of associated competencies in the Competence Ontology need to be specified. This
information is used by the AEHS to determine which competencies need to be covered for reaching a specific competence level. Consequently, the scope of Learner’s Knowledge Space is altered from describing learner’s state of knowledge to the description of learner’s existing competencies.
preferences stored in the User Model. In this work, we use this suitability function for weighting each connection of the Learning Paths Graph. From the weighted graph, we then select the most appropriate learning path for a specific learner (personalized learning path) by using a shortest path algorithm.
Figure1: AEHS architecture for lifelong competence development
As a result, the sequencing method of adaptive educational hypermedia needs also to be modified, to reflect the needs of generating personalized learning experiences for lifelong competence development. To this end, we propose an alternative sequencing method that instead of generating the learning path by populating the concept sequence with available learning resources, it first generates all possible sequences that match the desired competence level and then adaptively selects the appropriate personalized learning path from the set of available paths. More precisely, the following two steps procedure is used (as presented in Fig. 2): - Step1: Learning Paths Generation. At this step a graph containing all possible learning paths based on the relation between the Competence Development Programme, the elements of the Competence Ontology and the learning resources contained in the Media Space, is generated. - Step2: Personalized Learning Path Selection. At this step a personalized learning path is selected from the graph that contains all the available learning paths based on learner’s attributes in the User Model. As a result, we introduce an additional layer (see Fig. 2) in the abstract sequencing layers of adaptive educational hypermedia systems, namely the Learner Adaptation Layer, which is used for selecting the personalized learning path. In our previous work [6] we have proposed a decision-making function that estimates the suitability of a learning resource for a specific learner by relating the educational characteristics of learning resources defined in the educational resource description model with the learner’s cognitive characteristics and
Figure 2: Abstraction layers of AEH sequencing for lifelong competence development
3. Simulation Results and Discussion For the design of the Adaptation Model in our simulations, we have used the methodology presented in [6] based on a reference set of 30 learning object metadata records and a set of 5 simulated learner instances. These sets were used to calculate the suitability function discussed in section 2 of this paper. For the design of the Competence Domain Model we have created a Competence Development Programme and a relevant Competence Ontology (see Fig. 3) using the EuroPass Language Passport [7], a European common model for language competencies.
Figure. 3. Partial View of Competence Ontology of the Competence Domain Model in use For our simulations we created an additional set of learning object metadata records consisting of 12.500
records, with normal distribution over the value space of each metadata element. Additionally, we created a set of 50 simulated learner instances with normal distribution over the value space of each learner characteristic. These estimation sets were used for evaluating the efficiency of the proposed approach in generating appropriate learning paths. In order to assess the proposed approach, we compare the produced learning paths with those produced by a simulated ideal rule-based AEHS, using a specific Domain Model and Media Space. In our simulations we measure how close the learning paths produced by our proposed methodology are to these ideal paths for 30 different cases (10 randomly selected learner instances per Competence class). By this way we intent to demonstrate the capacity of the proposed methodology and investigate parameters that influence this performance. To this end, we have defined an evaluation criterion based on Kendall’s Tau, which measures the match between two learning object sequences, as follows: 1 N concordant − N discordant , k = + ∑n 2 k ( k − 1) ∀competence level
Success (%) = 100 *
where Nconcordant stands for the concordant pairs of learning objects and Ndiscordant stands for the discordant pairs when comparing the resulting learning objects sequence with the ideal reference one and n is the maximum requested number of learning objects per topic level for a targeted competence. Average Sequence Generation Success per Level of Sequence Root 100 95
%Success
90
be the resulted LO sequence, producing less probability of possible mismatches. Accordingly, for the same number of requested objects per topic, the more complex the targeted competence is, the longer would be the resulted sequence introducing more mismatches.
4. Conclusions In this paper, we address the problem of designing adaptive educational hypermedia systems for supporting lifelong competence development. To this end, we propose an alternative to the traditional subject domain specific design approach which is based on the use of competence description ontology and learner’s competence records. The feasibility of the proposed approach is verified through simulations under the assumption that appropriate metadata for describing competencies, as well as, learners’ competence records exist. In our simulations we compare the learning paths generated by the proposed methodology with those produced by a simulated ideal rule-based AEHS. The simulation results provide evidence that the proposed methodology can generate almost accurate learning paths for the attainment of complex skills and competencies by a given learner.
5. Acknowledgements The work presented in this paper is partially supported by European Community under the Information Society Technologies (IST) programme of the 6th FP for RTD project TenCompetence contract IST-027087.
maxLOs/Topic Level n=5
85
n=10 n=20
80
References
n=50
75 70
[1].
Field, J. Lifelong education. International Journal of Lifelong Education, vol. 20(1/2), (2001), pp. 3-15.
[2].
Aspin, D. N. and Chapman, J. D. Lifelong learning: concepts and conceptions. International Journal of Lifelong Education, vol. 19(1), (2000), pp. 2-19.
[3].
Friesen, N. and Anderson, T. Interaction for lifelong learning. British Journal of Educational Technology, vol. 35(6), (2004), pp. 679-687.
[4].
Brusilovsky, P. Adaptive hypermedia. Methods and techniques of adaptive hypermedia. International Journal of User Modeling and User-Adapted Interaction, vol. 11(1/2), (2001), pp. 87-110.
[5].
De Bra, P., Aroyo, L. and Cristea, A. Adaptive Webbased Educational Hypermedia. In Levene, M., Poulovassilis, A. (Eds.). Web Dynamics, Adaptive to Change in Content, Size, Topology and Use, Springer, (2004), pp. 387-410.
65 Complex Competencies Level 2
Complex Competencies Level 1
Simple Competencies
Competency Class in Competence Ontology
Figure 4: Average simulation results for learning path generation
Average evaluation results are shown in Fig. 4 presenting the success of the proposed sequencing method for different competence classes and different cases of maximum requested number of learning objects per topic level (n). From these results we conclude that the success rate of the resulting learning object sequences is depending on the complexity of a competency, as well as the maximum requested resources for each related topic. The less number of resources per topic are requested, the smallest would
[6].
Karampiperis, P. and Sampson, D. Adaptive Learning Resources Sequencing in Educational Hypermedia Systems. Educational Technology & Society Journal, vol. 8(4), (2005), pp. 128-147.
[7].
EC. Europass Language Passport, (2000). Available online at: http://europass.cedefop.eu.int