ROUGH DRAFT

ROUGH DRAFT - DIFFERENT USERS, DIFFERENT VOCABULARIES & DIFFERENT VIEWS

Paul Shabajee

[Editorial notes – this was written in a single pass with minimal retrospective editing, it needs heavy editing, rationalisation]

BACKGROUND

This is a generic set of issues revolving around the observation and requirement that different classes of user are likely to require different views on the same set of digital resources/assets/objects and where they are from different conceptual perspectives e.g. academic subject disciplines, levels of technical knowledge, etc., they are also likely to wish to locate and retrieve the assets using search interfaces and vocabularies appropriate to their background, capabilities and needs.

This is particularly the case where digital objects or components of composite digital objects (e.g. parts of a Learning Object) are potentially valuable across a range of applications and/or to a range of diverse users. As is likely to be the case in with respect to Learning Object and Visual Image Support use cases.

LEARNING OBJECT USE CASE AS AN EXAMPLE

If we take the Learning Object use case. In many models of an e-learning economy or eco-system, the re-use of often expensive and high quality resources/assets is seen as being a fundamental element. An active area of research (and debate) is to develop methodologies for the creation of Learning Objects that enables the maximal re-use of their components – the motivations being not only efficiency but also basic economic need.

The decoupling of components of a Learning Objects (e.g. external links and body text of an HTML page or animations and captions, Java-applet and data, …) is an important aspect design for re-use methodologies and debates. If this level of re-usability and thus granularity is desired then the demands and requirements on a digital repository for Learning Objects become richer e.g. requiring the rich description of sub-objects and means to locate and retrieve them.

There are two independent aspects here 1) a place to archive and retrieve, generally composite learning objects (e.g. Learning Object Content Packages) and 2) a repository of ‘raw’ components (of various levels of granularity) to support the creation of new composite Learning Objects. Both functions are requirements of any institution actively engaged in the development and use of Learning Objects.

[Needs a figure here showing the LOs and 'raw' objects and their relationships in the 'LO economy']

It seems likely that at a minimum, the level of integration of these functions, would require that they providing the functions are interoperable, i.e. provide interfaces utilising common standards, but it could be argued that a single repository capable of meeting both requirements is desirable and potentially more efficient.

However this would place yet greater requirements on the system e.g. the repository would become an institutional repository to store all fine-grained objects available in their most basic form e.g. original scan of an image, which will then be edited for use in specific Learning Objects. Each new version of the object would then be held in the system and would require some version tracking and associated metadata to describe the relation between the versions.

IS THIS RELEVANT?

Thus IF part of the Learning Object use case, is to support the authoring of Learning Objects by enabling authors to locate and retrieve potential resources/assets from the repository for use as components of their own new learning objects, then the repository will need to expose metadata which will support those activities as indicated in McLean and Lynch (2003) in their Draft paper on ‘Interoperability between Information and Learning Environments – Bridging the Gap’ p8. A related example is students from different disciplines using the repository as a source of objects to use within in assignments.

IF these (and other) activities are to be supported AND users of the systems are likely to be from diverse subject disciplines the issues of providing appropriate discovery/descriptive metadata become yet more problematic than a use case where only members of the ‘depositing’ communities are likely to wish to retrieve the object.

OTHER PROBLEMS AND SOME SOLUTIONS

As detailed in Shabajee 2002, describing fine-grained resources/assets, such as still images, for multi-disciplinary access is a very problematic and yet vital element of any system wishing to support users from multiple disciplines. These issues are related to the discussion of ‘heterogeneous schema and instance data support’ section of this [SIMILE Use Case] document, however there is the added set of problems of finding ways to describe ALL resources where cross-disciplinary uses are likely (i.e. the resource may be of value to multiple disciplines), with appropriate descriptive metadata vocabularies from multiple communities. There are many ‘hard’ and fundamental problems involved e.g.

The size and number of potential specialist community vocabularies is vast – how can the practical need to limit the number of terms be balanced against impact limiting access to potentially valuable resources – i.e. if you can’t find a resource when you need it, in the context that you need it in, it is to all intents and purposes, not there.

Members of ‘depositing’ communities are not likely to be able to apply indexing terms from other communities as appropriate – e.g. they would be unlikely to have the necessary expertise or indeed time or motivation.

It is likely that communities vocabularies/ontologies will change, how are changes to be retrospectively applied etc…

...

There are many possible solutions to this set of problems [multiple disciplinary markup] e.g. provide semi-automatic indexing tools using concept extraction/textual analysis tools, create mappings (where possible) between domain vocabularies/ontologies allowing simple mappings and inferencing about relationships between terms, provide a high level vocabulary that provides easier/less technical although less fine grained descriptions, Content Based Image Retrieval (CBIR) systems…

A further non-technical solution is to employ indexing/metadata specialist(s) [in house or externally] who do the indexing or validate and edit the indexing of original contributors. In much the same way that in many Universities the function of obtaining and validating copyright clearance for media/assets used in Learning Objects is a centralised or federated function.

This latter example highlights the critical Rights Management issues. In many cases components of Learning Objects the copyright belongs to a 3^rd party, in some cases significant use restrictions are likely to be imposed by the copyright holder and these must be tracked and access restricted on the basis of these – which have particular issues in the archiving and long term reuse (even in the original form) of composite Learning Objects.

It should also be borne in mind that users will be searching for resources from multiple sources not, in general, simply from a single repository, thus an authoring system that interacts with repositories should ideally provide cross searching facilities and the individual repositories need to provide interfaces to allow this… Again see McLean and Lynch (2003)

The issues above have been distinct from any User Interface (UI) differences in providing access to different classes of users [or individuals]. However the design of the UI will interact with these factors e.g. functional role, background, capabilities and needs of the user, e.g. at a macro level – tutor, student, learning object designer, librarian, copyright clearance admin… will require different views on the data and metadata held by the system.

At a finer grain, [for example] students or lecturers of different subjects and levels, will require different views and interfaces e.g. browsing a collection of objects using their own community ‘ontology’ using hierarchical tree views containing only terms relevant to them. Specific UI’s will be required to provide views of data and metadata that meet their conceptual and functional tasks… This is obviously tightly tied into the work and goals of the Haystack project.

A related issue is that of machines and agents as users – once again depending on their ‘needs’ descriptive vocabularies and interfaces will need to be customised/able?

VISUAL IMAGE SUPPORT USE CASE

Although the foregoing discussion has been based around the Learning Object use case, analogous arguments run true for Visual Image use case, indeed in many ways the Visual Images use case can be seen as a sub-set of the Learning Object use case IF one of the requirements of the Learning Object use case is to support fine grained Learning Object archiving and access as described above.

REFERENCES

McLean, Neil and Lynch, Clifford. (2003) Interoperability between Information and Learning Environments – Bridging the Gaps, A Joint White Paper on behalf of the IMS Global Learning Consortium and the Coalition for Networked Information. DRAFT – Version of June 28, 2003 - http://www.imsglobal.org/DLims_white_paper_publicdraft_1.pdf

Shabajee, P. (2002) 'Educational Metadata' a Fundamental Dilemma for Developers of Multimedia Archives, D-Lib Magazine, Vol 8(6) Available: http://www.dlib.org/dlib/june02/shabajee/06shabajee.html