19th December 2002, Milton Keynes

The following report is from the 19th December 2002 Metadata & Digital Repository SIG Meeting, held in Milton Keynes. Report by Phil Barker. Presentations are available to download, which we hope you find useful.

Welcome, background to the SIG
Phil gave an introduction to the aims and activities of CETIS and the metadata and digital repository interoperability SIG. [powerpoint presentation, (45kb)]

MDR Specifications and Standards Update
The vote on whether to accept the draft Digital Repository Interoperability specification will take place between 6 Jan and 17 Jan 2003.

The IEEE are currently active in developing bindings using XML Schema (and RDF). [As was pointed out by Eddie Clarke later in the day, the IEEE 1484.12.1-2002 standard (the LOM) covers the information model only, there is currently no standard binding.]

Phil advised IMS are involved in some "maintenance" activity for the Meta-data specification.

Lorna added that the LOM is not likely to become an ISO standard in the near future.

The CORES forum is addressing issues relevant to some of the problems encountered with XML bindings of the LOM/IMS Metadata by suggesting that every metadata element is assigned its own URI which will allow more flexible schemas to be used.

IMS have changed their structure; the new structure allows "charters" under which new development work on specs can be undertaken. Work on the meta-data spec. may start up again.

MEG Update (Pete Johnston)
MEG is a wide ranging group of people interested in the use of metadata in the service of education, which can be joined via the email list, taking part in discussions on that list and at meetings. Also, to register your support for MEG's objectives of promoting interoperability standards, it is possible to sign up to the MEG concord-some 80 organisations have done so thus far.

Other MEG activities are:
 * development of a common scheme for describing educational level;
 * liaising with the office of the e-envoy;
 * the development of an application profile register, a workshop on which will be held in Bath on 21 Jan 2002.

MEG and the CETIS MDR SIG cover similar areas, MEG is somewhat wider in that it includes schools and lifelong learners, community education, museums etc as well as HE and FE, and also that it offers a place to discuss the use of resources which were not specifically designed for education (as found in, e.g., museums). MEG and the MDR SIG are looking at ways of working more closely together, one suggestion is that we hold back-to-back meetings on consecutive days.

IMS DRI specification
Andy Powell gave an overview of the IMS DRI specification, describing what digital repositories are and the "interoperations" which can be supported by the spec. He compared this to the technical architecture of the Jisc Information Environment, which is similar. [Andy is the author of this architecture, he also gave a presentation to the nascent IMS DRI working group, but was not involved in the development of the spec.] The Jisc's DiVLE programme and the Jisc X4L JORUM+ project will both be test beds for digital repository interoperability (see the Jisc website, www.jisc.ac.uk, for more details on these).

In discussion a few areas which are not in the scope of the specification were mentioned which are worth noting here: recommendations based on profiles of users of similar resources (a la Amazon); commerce issues, eg payment; identification of which repositories to search, ie collection level descriptions (UDDI was mentioned in this context).

For more details, see Andy's presentation on the UKOLN site.

CETIS MDR SIG Metadata tasks update
The status of current SIG tasks is as follows:
 * The website is being moved to http://metadata.cetis.ac.uk
 * Jenny's introductory guide to metadata was circulated as her final draft on the jiscmail list and will be published by CETIS shortly
 * The survey of educational metadata use has spawned a series of case studies; a draft of the first of these has been circulated on the jiscmail list.

Comments on the above are welcome. For more details, see Phil's slides [50KB].

Suggestions for new tasks were requested. One suggestion from before the meeting was a comparison of metadata from different sources, e.g. information professionals, users (teachers, learners), creators. Others suggested during the course of the meeting were:
 * A detailed guide to the LOM, developed through discussion on the jiscmail list
 * A forum / register of projects and people with expertise in metadata [feel free to use the sig jiscmail list for this purpose, though this doesn't preclude us developing something more structured]
 * Information on the tools available for metadata creation, storage and management
 * A critique of the educational elements in the LOM and possible alternative approaches.

Phil will follow up these suggestions on the JISCmail list.

OAI and IMS Metadata
Steven Richardson showed us how he had been able, with relatively little effort, to add OAI-PMH support to a relational database which had been designed using a schema which was related to the LOM. He has extended this support so that the Results OAI service will provide IMS records as well as the simple DC records mandated by OAI. [The major effort was in understanding the LOM, that done the OAI support was added in about a week and the extension to IMS records in another week.]

Steven's presentation includes details of the third-party scripts he started with, and he hopes to be able to make the modified versions of these scripts available. A paper by Steven and Andy Powell has been published in Ariadne Issue 34.

Andy Powell pointed out that while Steven describes sets OAI records being generated on the fly by a query to the SQL database, OAI-PMH is not a protocol for passing search queries: the OAI sets have to be pre-defined.

Steve Jeyes pointed out that it is difficult to interoperate when you don't have anyone to take the records you are exposing.

For more details, see Steve's slides [128KB].

EnCoRe
Melanie Keady presented on the EnCoRe (Enriching Courses with Resources) project, which is funded through Jisc's DiVLE programme. She described the project and its aims and invited comment on the issues facing the project, especially the choice of standards/specs (DC, SCORM, OpenURL), metadata/content package editors, and taxonomy.

There was discussion about the pros and cons of SCORM: on the one hand it is not designed to support much of the learning activity relevant to UK HE (it won't support IMS Learning Design), on the other hand it is stable and working.

Andy Powell advised that DC-Dot was not designed for the type of metadata editing which the EnCoRe project was involved in. Lorna advised in favour of the use of LRN (it works reasonably well, see the content sigs report on ...) while EC-pac is being updated.

Lorna also advised using Dewey for subject classification, though there were questions about the cost of this. The BEI Thesaurus was also mentioned as a possibility (see http://brs.leeds.ac.uk/~eclts/beir.html).

For more details, see Melanie's slides [310KB].

RESL
David Massey presented an introduction to RESL, the re-usable educational software library, which is an OU-based project building a library of resources to support good practice in the reuse of educational software in HE. The resources in the RESL library include information about the use of educational software and the software itself. The IMS metadata schema has been used as a basis of the RESL descriptions, in particular keyword vocabularies were developed for subject, pedagogy, strategy and technology and these were encoded in the Classification section of the IMS schema (see http://kn.open.ac.uk/public/document.cfm?documentid=2579). David also gave us a demonstration of searching and browsing the RESL catalogue - the browse functionality is based on the keyword vocabularies mentioned above.

For more details, see David's slides [889KB].

Paul Shabajee (ARKive-ERA)
Paul Shabajee gave a presentation covering similar ground to his recent D-Lib paper "Primary multimedia objects and 'educational metadata': a fundamental dilemma for developers of multimedia archives" (June 2002, D-Lib Vol8 No6). This is based on his experience with the ARKive-ERA project which is concerned with the educational repurposing of assets such as images and movies.

The dilemma he describes is that if a resource cannot be found it might as well not exist, and in order to find resources we need metadata; however creating a metadata description can limit the people who will find the resource. For example, no one cataloguing images of birds would encode the information that birds fly, it would be taken for granted, so anyone searching for information about flight might find images of flying fish and flying squirrels, but not of birds. The context of use determines what is significant and what is trivial. Another example is that many portraits show people wearing hats which might be useful for a history of fashion course, however most cataloguers would describe the person, not the hat. Thus for the purposes of someone interested in the history of fashion these images might as well not exist. Paul's Catch-22 dilemma is:

You don't want to (and can't) predict what your users will want to use the 'raw' multimedia assets for, but if you don't, your users can't get to the assets.

For more details, see Paul's slides [1MB].

COSE
Eddie Clarke spoke about the use of metadata in COSE-the Creation of Study Environments VLE which is characterised by a pedagogy based around the re-use of content for learning activities. COSE 2.05 uses version 1p2p2 of the IMS metadata specification, and Eddie drew the distinction between being instance conformant and system conformant (COSE like most other VLEs is instance conformant, it can read in the metadata which it requires from a content package). How this metadata is used in COSE is set out in a report which Eddie referenced (appendix F of the COSE manual).

COSE is involved in the SURF X4L project which will drive the further use of metadata since it is focused on the exchange of content via repositories.

Among the issues which Eddie raised are lack of consensus on conformance requirements, the need for agreement extensions and vocabularies so that interoperability is not compromised, the relationship with SCORM, and linking metadata to web pages in the way which is possible for DC.

For more details, see Eddie's slides [38KB].

Education Elements in the LOM
''At a recent CETIS Content SIG meeting, Steve Jeyes (CETIS) had made a few comments suggesting that he had a low opinion of the Education elements in the LOM-he is not the only person I have heard voice this opinion. I think that any problem with describing the educational aspects of a resource is a problem at the heart of a schema specifically aimed at the description of learning resources, and so I invited him to put these criticisms up for further discussion at this SIG meeting.-Phil Baker, SIG coordinator''

Steve started by showing that many of the metadata schema which have been based on the LOM and its drafts have chosen not to use many of the education elements, which suggests a problem. He went on to ask what one would want from metadata when choosing an educational resource, eg:
 * Information and criticism, Steve drew a comparison with choosing a wine "if it is Merlot [information] and Gluck rates it highly at 16/20 [critique] I'll buy it".
 * Dynamic control, Steve envisaged a student being able to say to their VLE "this resource is the right subject but I want something easier or harder " [or more visual, or more active].
 * Level, eg it is necessary to specify whether a resource is a single object or an entire course.

These are just examples: there is a need to develop a set of relevant use cases to ensure metadata can support real user needs.

Steve then went through each of the elements in the education section of the LOM and outlined some of the criticisms of them (plenty of people joined in with this). For example: the vocabulary for the learning resource type is a bit of a mish-mash terms which seem to be more relevant to books than electronic resources-there have been various attempts at improving on this; the education context is not rich enough for use within a given sector; the difficulty only has meaning if the context is well described. It was also commented that, as Paul Shabajee had shown earlier, not all education elements can be applied to all resources: difficulty and sematic density can't easily be applied to images.

Steve pointed to Tom Carey's "Educational rationale metadata" as a way of encoding the intention of the designers of learning resources. A complementary approach is the use of relevant "purposes" in the "classification" section (as RESL did).

For more details, see Steve's slides [120KB].

''This was a lively and much interrupted talk, with some of the comments abovecoming from comments by other people rather than from Steve. I don't think anyone disagreed with Steve's assertion that the education section is not the strongest part of the LOM. However it is not enough to point out a failing: in order to progress we need find something better. Few people will use all the education elements, and if we are to pick and choose then it would be beneficial if we pick and choose the same elements. Consensus building is even more important if extensions to the LOM are to have value for interoperability. Hopefully, we will be able to continue discussion and consensus building through the SIG's other forums and activities.''