Achieving Learning Object Repository interoperability

Position paper for the LIFE1 workshop on Learning Object Repository Interoperability

Monday 5 September 2005, Manchester, UK

        Fredrik Paulsson
        KTH (Royal Institute of Technology)
        & National Agency for School Improvement
        frepa@nada.kth.se | fredrik.paulsson@skolutveckling.se

Current situation
The issue of Learning Objects Repository interoperability rests heavily on the exchange and interoperability of learning objects metadata. The focus on metadata has lead to a considerable legacy from Library and Information Science, where archive interoperability and metadata exchange have been important issues for many years. This legacy is however a double-edged sword: on one hand there is a lot of experience and knowledge to acquire, on the other hand the metadata requirements and types of repositories are quite different as well as the conception of the metadata concept. However in recent years research and development on Learning Object Repositories (LOR) have started to distinguish it self and shape an agenda of it own. This is much due to the acceptance and maturity of metadata standards such as LOM and recently to some extent due to development of interfaces and standards for the exchange of metadata – such as the SQI.

LOR “hubs” are slowly starting to appear around Europe (and the rest of the world as well) as well as related brokerage services for Learning Objects. Those hubs are typically working according to one (or several) of three main principles spanning from basic metadata exchange through a central metadata repository and/or “metadata harvesting”, through federation of metadata queries, to the use of Peer-to-Peer technology (P2P) and loosely connected LOR-networks – all with their respective pros and cons.

Key issues
Open standards
The basis for large scale interoperability is an extensive use of open standards. Open standards for, especially metadata, have gained a wide spread acceptance during the last 5-10 years. There are however still a lot to do and it is important to promote open standards for all layers of the learning (and learning objects repositories) architecture. Currently there is an urgent need for a higher degree of consensus on standards for query interfaces and query languages.

At the same time it is important to provide mechanisms for co-existence of different standards for different needs and ambitions. Emerging standards must therefore support mechanisms for determining which sets of standard is supported by a specific LOR in terms of query language, Application Profiles etc.

Metadata quality
Irrespective of the used approach for metadata exchange, one of the most (if not the most) critical issue is the metadata quality. In most cases where two or more Learning Object Repositories are exchanging metadata the metadata models are “forced” to harmonize. This is often done by extracting the smallest common set of metadata which causes the metadata model to be transformed from an often fairly primitive metadata model to a very primitive metadata model. As the quality of metadata drops, the amount of redundant information returned from queries increases rapidly and the usability of the service drops substantially – especially as a part of a brokerage system where the overall amount of metadata tends to get very extensive. Our experiments using federated metadata queries show that, in some cases when large repositories are involved, the system has to deal with several Mb of irrelevant metadata before the final search response can be delivered. This also means that any additional metadata filtering on the “metadata hub” has to wait for all repositories to return answers or time out. One of the key issues concerning metadata quality is metadata semantics and semantic interoperability.

The problems with metadata quality could be addressed by setting up objectives such as:

- making better use of Semantic Web technologies for Learning Object metadata and, [2] [3]
- By a more extensive use of application profiles. [1]

Together they likely to increase the quality of metadata and metadata semantics and in the same time provide mechanisms for mixing different metadata models and schemas as well as enhancing semantic interoperability – which will ultimately lead to a higher degree of Learning Object repository interoperability. [1]

In addition there is a need to drop some of the heritage from library and information science regarding the view of metadata and move on from regarding metadata as something static, owned and managed by a few, to being a dynamic, growing echo-system where many different actors provides their pieces of the puzzle based on their angle and relevant to their specific context. Such view emphasises the need for more sophisticated mechanisms for Learning Objects Repository interaction (such as P2P). Such metadata view might even change the conception of what a LOR is to embrace other type of information spaces such as portfolios, local metadata stores, blogs etc. Such development is likely to increase the need for loosely connected networks with a high level of flexibility.

Query language
As the metadata quality increases and the semantics improve it will be necessary to make use of query languages that supports richer metadata semantics. This is most of all essential in order to allow for metadata filtering at “LOR -level” in order to avoid huge amounts of redundant metadata - even though the metadata might be of good quality. [7] This is especially important with a P2P approach. [4]

Recommendations

References

[1] Heery R, P. M. (2000). Application profiles: mixing and matching metadata schemas - introduce the 'application profile' as a type of metadata schema. Arriadne(25).

[2] Naeve, A. (2005). The Human Semantic Web – Shifting from Knowledge Push to Knowledge Pull. International Journal of Semantic Web and Information Systems (IJSWIS), 1(3), 1-30.

[3] Nilsson, M., Palmér, M., & Naeve, A. (2002). Semantic Web Meta-data for e-Learning - Some Architectural Guidelines. Paper presented at the 11th World Wide Web Conference (WWW2002), Hawaii, USA.

[4] Nejdl, W., Wolf, B., Qu, C., Decker, S., Sintek, M., Naeve, A., et al. (2002). Edutella: A P2P Networking Infrastructure Based on RDF. Paper presented at the 11th World Wide Web Conference (WWW2002), Hawaii, USA.

[5] Paulsson, F., & Engman, J. (2005, 19 - 21 October 2005). Marking the National Curriculum - a new model for semantic mark-up. To be presented at the eChallenge, Ljubljana, Slovenia.

[6] Palmér, M., Naeve, A., & Paulsson, F. (2004). The SCAM Framework: Helping Semantic Web Applications to Store and Access Metadata. Paper presented at the European Semantic Web Symposium 2004, Heraclion Greece.

[7] Henze, N., Maluszynski, J., & Bry, F. (2003). Principles and practice of Semantic Web reasoning: international workshop, PPSWR 2003, Mumbai, India, December 8, 2003: proceedings. New York: Springer.

[8] Kraan, W. (2003, November 05, 2003). Using SHAME to fill your SCAM. Retrieved May 12, 2004, from http://www.cetis.ac.uk/content2/20031105152216

[9] Paulsson, F. (2003). Standardized Content Archive Management – SCAM. IEEE Learning Technology newsletter, 5(1), 40-42.

[10] Palmér, M., & Naeve, A. (2005, July 18-22). Conzilla – a Conceptual Interface to the Semantic Web. Paper presented at the Invited paper at the13:th International Conference on Conceptual Structures, Kassel.

------------------------------

[1]  The LIFE project is partially funded by the European Commission. The project is led by The European Schoolnet with the following partners:University of Vigo, CETIS (University of Bolton), University of Oslo, University of Paris X, European IMS Network, SUN Microsystems, and CEDEFOP

--------------------

[This page was last updated on 2005-08-31 by Fredrik Paulsson]