17th August 2004, Glasgow

The following report is from the 17th August 2006 joint meeting of the CETIS Metadata and Digital Repository SIG and the Metadata for Education Group, held in Glasgow, on the UK LOM Core. Report by Phil Barker, based in part on notes taken by Lorna Campbell.

A list of attendees is given at the end of this report.

Introduction
Most of the agenda for this meeting was based on issues that had been raised on the CETIS Metadata JISCmail discussion list during July and August. The JISCmail discussion of each issue was summarised in an email circulated before the meeting which provided the starting point for discussion at the meeting; links to these emails are available at the appropriate point in these notes. The remainder of the agenda covered possible future developments for the UK LOM Core.

About the UK LOM Core
Lorna Campbell, one of the authors of the UK LOM Core, gave a brief introduction to the UK LOM Core, where it came from, who developed it and why.

The UK LOM Core started out as a research oriented task undertaken by Lorna and Gerry Graham of LT Scotland, who carried out a survey to establish which LOM elements were most widely used in UK Education. It was then used as the basis of advice on which elements should be used by projects in UK Education wishing to share their metadata, and at this point started to include advice on how those elements should be interpreted. It was (and still is) expected that projects and services would create an application profiles based on the UK LOM Core rather than use the UK LOM Core as it stands.

Lorna noted that since the UK LOM Core had grown from a survey of existing practice into advice for encouraging good practice in a fairly ad hoc manner, at no time were use cases collected. Many of the problems to be discussed at this meeting might be resolved by collecting use cases for the UK LOM Core.

Issues arising from the second draft
The notes I made on the day don't always make it clear who said what; sorry about the unattributed comments which result from this.

What is the UK LOM Core for?
The introduction to draft 2 of the UK LOM Core states why the work was started, not what it is for. What kind of resources is it for? (It has been used for assessment items, web sites and courses.) Is it to facilitate technical interoperability? Is it to facilitate the discovery of resources for teaching and learning? If it is to cover a wide range of possible uses does it need to be more flexible, allowing different levels of implementation?

Sarah Currier and Jane Barton felt that the focus of the UK LOM Core should be on exchange of metadata, not on how projects should handle their internal metadata workflow. Interoperability does not rely on everyone using the same standard, it only requires that when you're describing the same characteristics they can be exchanged. If the LOM doesn't meet the needs of a particular project they can use something else, but still be able to exchange metadata with other projects using the LOM. The global needs met by the LOM should not interfere with the specific requirements of a project.

In this way problems with describing specific resources (e.g. questions, physical resources) need to be dealt with by projects but are not necessarily the concern of the UK LOM Core.

Barry Kruger: It is important to manage people's expectations, saying it will facilitate interoperability could be dangerous. People might think that they can create a minimal UK LOM Core record and then be able to submit that record to Curriculum Online, NLN etc. It would be more appropriate to say it will predicate interoperability but will not necessarily guarantee it.

Jane Barton suggested we clarify what the UK LOM Core is for from service and data providers' points of view.

Mike Collett asked whether there would be a conformance statement, and Ben Ryan said that conformance statements needed to be built in right from the start. John Bell offered to profile UK Lom Core in phase 2 of Telcert (starting in January) so that conformance tests can be built. [See also Ben Ryan's presentation on XML bindings and validation below]

For the next version of the UK LOM Core:
 * Rewrite introduction to clarify what UK LOM Core is for (and not for), rather than how it was developed.
 * Identify the kind of resources UK LOM Core is applicable to.
 * Explain that the profile is designed to facilitate exchange and interoperability not internal workflow process.
 * Define interoperability in this context.

For more details, see the JISCmail discussion summary.

Mandatory elements
The summary suggests that there may be a need multiple / optional sets of mandatory elements and different levels of conformance for different types of resources of different project needs. John Bell gave an example of a project in Mexico using HarvestRoad which took this approach.

Jane Barton said that the UK LOM Core should facilitate semantic interoperability, whether an element and its value are there or not should be determined by the application. However, Barry Kruger said that SMEs want to be told which elements and vocabularies they should be using.

The beneficiaries of the UK LOM Core are (hopefully) the user community who need to be able to find resources: what do they need by way of metadata elements that must be present to search on, and what will cause them problems by way of requiring that semantically nonsensical metadata is generated for a mandatory field? What is the loosest community of practice for which you can define mandatory elements? Really need use cases to answer these questions.

Several people suggested variations on the theme of having two ways of applying the UK LOM Core: content Vs structural or semantic Vs syntactic; i.e. conforming to the element guidelines and definitions and producing a record that conformed to the mandatory elements.

For the next version of the UK LOM Core:
 * Divide UK LOM Core into two parts: guidelines, and profile of mandatory and recommended elements.
 * Re-organise document to reflect this change.
 * Conformance statement: implementers may comply with guidelines or conform to profile.
 * Status of all elements remains unchanged.

For more details, see the JISCmail discussion summary.

Null values
Ben Ryan said that if you had mandatory values you needed to allow null values; the only question was what the null value for each mandatory element should be.

For the next version of the UK LOM Core:
 * Null values are acceptable for all implementations conforming to the profile.
 * Null values for mandatory elements must be reviewed on a case by case basis.

For more details, see the JISCmail discussion summary.

vCard
All agreed with the solution proposed by Chris Hubick and summarised by Andy Powell. Ben Ryan commented that the lack of an XML binding for vCards was a real problem.

For the next version of the UK LOM Core:
 * Update vCard guidelines to reflect the solution proposed by Chris Hubick.

For more details, see the JISCmail discussion summary.

Rights category
End users express a wish that this information be present to inform their choice of resources, however to be useful the information needs to be accurate and up to date.

Rob Tice said that for Curriculum Online, a set of extensions had been provided to meet the need of describing restrictions placed on the use of the resource. Rowin Cross pointed out that in the IMS QTI 2 specification was ambivalent about the presence of rights metadata.

Charles Duncan suggested that a reference to a rights statement in, e.g. ODRL, would be useful, but that since the rights.description element was a langstring not a URI, it seemed that this was not the place to put it. Charles also suggested that the addition of a role of "Rights Holder" to the lifecycle.contributer element would solve many problems.

There was a fair amount of discussion over the interpretation of the vocabulary ('yes', 'no') provided by the LOM for elements rights.cost and rights.copyright and other restrictions, and whether the latter should be read as referring to the existence of copyright or copyright restrictions.

For more details, see the JISCmail discussion summary.

The issue remained unresolved.

For the next version of UK LOM Core:
 * Record issue as 'unresolved'.
 * Update guidelines to clarify that UK LOM Core assumes standard UK copyright applies by default.
 * Maintain status quo but make 6.3 conditional on either 6.1 or 6.2 being 'yes'.
 * Add the term 'rights holder' to lifecycle.contribute vocabulary.
 * Add appendix with Curriculum Online and CELEBRATE rights extensions.

Other comments
Sarah Currier suggested that the guidelines in the UK LOM Core should not contain assertations about who should create the metadata as, for example, is the case with element 1.5 general.keywords in the current draft.

Other comments are welcome by email, it was suggested that a "working copy" of the latest draft document be made available in order to avoid duplication of comments.

XML binding and instance validation
Ben Ryan gave a presentation describing approaches to enabling the validation of UK LOM Core instances. He has created a W3C XML Schema for the UK LOM Core and will produce a RelaxNG schema. Validating instances bound with these schema requires that one check that the XML is well-formed, that the structure conforms to the schema, that the content is valid and that the correct vocabulary is used. The possible tools for validation are DTDs or Schema, programmatic validation, XSLTs and Schematron. DTDs lack support for some advanced constructs, and Schema lack support for XML entities; programmatic validation requires advanced skills.

The approach Ben has taken is to perform the basic structural validation using DTDs and schema, to use Schematron for more advanced structural validation and to use XSLTs for the content and vocabulary validation. Schematron is designed to validate XML content using rules that can be either mandatory or recommended. Having vocabularies specified in VDEX along with some simple conventions, allows scripts using XSLT to compare the vocabulary used in XML instances to be compared with those in the vocabulary file. This approach supports phased validation: one can define what is required at different stages of the metadata creation workflow.

Ben will make all the DTDs, schemas, applications etc. which he has used available to others who wish to take a similar approach. Ben and TELCERT are to liaise.

Becta update
Barry Kruger gave a brief update on developments at Becta, notably concerning version 3 of their tagging tool. This version is modular and uses config files to define the application profile being used; validation rules are specified using XML schema, with separate vocabulary validation.

UK LOM Core and BSI
Mike Collett, who is chair of BSI IST/43, gave an overview of how a document such as the UK LOM Core can be taken forward through the BSI and what the pros and cons of doing so are. The BSI publish standards, guidelines and "publicly accessible specifications" (i.e. work from outwith the BSI). The advantages of doing so are that the document carries more weight, it is more likely to be adopted by government agencies and BSI will cover the maintenance. The disadvantages are that the BSI process could change the work and, while one doesn't have to pay to use the standard, BSI would charge for paper copies of the standard.

It should be possible, if we desired, to create a multipart standard from the UK LOM Core. Since small standards are considered better than monolithic behemoths, a viable approach might be to standardise stable vocabularies such as the UKEL first.

What next: additional actions
Lorna will publish the next version of the UK LOM Core by the end of December 2004.

A running draft of version 0.2 will be posted so that the community can view known issues and typos and so avoid duplication.

Attendees
Many people expressed their regret that they were unable to come because of the timing of the meeting.

Many thanks to all who attended, especially to those who gave presentations and to those at LT Scotland who helped with the organisation.