Metadata and Digital Repositories

CETIS MDR SIG Meeting on Identifiers for Learning Objects

Report by Phil Barker of meeting held 1 July 2003, Bristol.

Agenda

1. Introductions
2. Background
3. Agree scope and objectives
4. Paper 1: Charles Duncan
5. Paper 2: TSO
6. Paper 3: Andy Powell
7. Next steps.

1 Attendees

2 Background

Phil Barker spoke about CETIS, and its role in supporting educational technology interoperability standards, and the remit of the metadata SIG. Identifiers fall into this remit since the IEEE LOM specifies that they should be used to identify the resource which the metadata describes, and that they should be used to identify metadata records (i.e. LOM elements 1.1:General.Identifier and 3.1:­Metametadata.­Identifier). We also noted that whereas previous drafts of the LOM (including that used as IMS Learning Resource Metadata 1.2.2) had reserved a single simple element for these identifiers, the final version had adapted the catalogue element of previous versions. So the IEEE LOM identifier elements are broken down into two sub-elements: one which gives the catalogue or identifier scheme and the second which gives the entry in the catalogue or the identifier.

It was also noted that IMS have picked up on identifiers for learning objects as an area for further discussion. Neil McLean of IMS Australia will be visiting the UK in the Autumn and a meeting will be held then to discuss the issue further.

3 Scope and Objectives

Phil B. noted that one of the complicating factors for these discussions is that identifiers have a much wider field of use than that covered by CETIS. On the other hand any scheme used for learning objects in UK F&HE would have to be compatible with a wider area of applicability, e.g. the e-Government Interoperability Framework. While the IEEE LOM is important to CETIS's field of work, any identifier scheme we propose would also has to work with other metadata schemas, e.g. Dublin Core.

Robin Wilson noted that TSO have a commercial interest in the wider governmental use of identifiers. He commented that it was useful to distinguish between identifiers per se (i.e. strings which are unique with some scheme) and the provision and maintenance of a service which makes sure that these are allocated each to a single object and provides a mechanism for resolving the location of an object (or other information relating to the object) when given the identifier. The identifiers them selves are cheap, however the maintenance and resolution services are expensive. This meeting mostly concentrated on the features of an identifier scheme (while recognising that availability of a resolving service is important in realising many of the benefits which follow from adopting a suitable scheme).

We decided that we would most likely be able to influence the wider discussion on identifier schemes if we could show progress in a more specific application. Given the interests of the people at the meeting, the remit of the CETIS metadata SIG, and our understanding of the current needs of the UK F&HE community it seemed appropriate that we limit the scope of our discussions to the IEEE LOM as used by the X4L programme[1], RDN/LTSN interoperability projects[2] and possibly the Learning and Teaching Portal [3]--or, in slightly more general terms, the application profile currently known as the UKCMF[4]. It was hoped that we could make significant progress to proposing an identification scheme for use within this context--or at least, if not a single scheme, then options which would work with each other (we agreed that any identification scheme would have to coexist with others without causing conflict).

The desire was also expressed that through this work the benefits of unique identifiers could be made clear.

In the papers below, the first discusses how identifiers are used, the next two discuss what identifier scheme may be used.

4 Charles Duncan's Paper

Available at http://metadata.cetis.ac.uk/sig_meetings/cd-identifiers.doc

Please note that this paper is based on a draft discussion document which pre-dates the adoption of Catalogue Entries to replace a simple Identifier element, and hence aims to make distinctions between the two. The aim of the paper is to show what the use of suitable identifiers allows us to do; that is identify a resource, identify copies of a resource, facilitate the location of a resource, identify different versions of a resource and identify the use of a resource in aggregations.

There are issues over who can create an identifier which cannot be changed, who can create an identifier which can be changed (or added to), and changes to the metadata as a result of aggregation or changes of location. We agreed that is was essential that each identifier referred to only one resource, and that it was desirable that, within a given scheme, each resource had only one identifier.

We spent some time working through representations of what would happen to the resource identifier, the metadata and the metadata identifier if a resource was created, uploaded to a repository, moved to another repository etc. We agreed that more example life cycles represented graphically would be a good way of communicating out thoughts on how identifiers could be useful--Phil to create examples based on the use cases in Charles's paper.

We also discussed the potential and problems presented by section 7 of the LOM which is used to store information on relationships. For example there is a relationship between a learning object which is an aggregation of other learning objects and the objects which are part of that aggregation. These relationships are two way: in the example given there would be a "hasPart" relationship in the metadata record of the larger aggregate resource and an "isPartOf" relationship in the metadata record of the smaller resource. The ability to identify popular resources, i.e. those with many "isPartOf" relationships could be very useful, however populating both ends of the relationship in a distributed environment could be problematic. Consistent use of identifiers would help enable a "harvesting" approach to solving this problem.

TSO White Paper on DOIs

The "Digital" in DOI stands for digital identifier, the object being identified need not be digital.

Rob Wilson introduced us to GURP as criteria for classifying identification schemes. GURP is an acronym for Global, Unique, Resolvable, Persistent/Permanent. The TSO white paper asks whether an identifier scheme with all four elements of GURP would create a new level of interoperability, linking resources to information on rights management etc.

The TSO run a resolution service for DOIs and are hopeful that this will be useful in applications such as rights management. Currently, in order to use DOIs for linking in a web environment a plug-in is required; it is possible that Microsoft might include native support for DOIs in a later version of Internet Explorer. For this to work the resolution service holds metadata on the resource, such as its location.

DOIs are based on Handles[5], the difference between the two is largely that the International DOI Foundation [6] manages a system for allocation and resolution of DOIs. As mentioned above, these services are the more costly aspects of identifier systems. The metadata used by these identifier systems, and the relationship between DOI metadata and other metadata schema, is described in the Indecs Data Dictionary [7].

6 Andy Powell's Paper

Available at http://www.ukoln.ac.uk/distributed-systems/lo-identifiers/

Andy's paper seeks to put forward the requirements of an identifier scheme for use within the Jisc Information Environment. Of the "GURP" criteria, only Global is missing (although this is implied by context and by the requirement that the identifiers should be transportable). The feeling of those round the table was that, for the scope of discussion at this meeting, persistence over a time scale of 10-15 years was more important than permanence. There was most debate over the proposed requirement that identifiers be "usable in web browsers"--i.e. that identifiers can be used in the same way that URLs are currently used. Clicking on an identifier link, or entering the identifier in the location bar could (via a resolver service) bring up the resource itself, check the rights for using the resource, or bring up a metadata record for the resource. It was argued that a non-actionable identifier could be sent to an OpenURL resolver to achieve this. It was also argued that the proposed requirement of identifiers being URI compliant was perhaps a way of meeting other requirements rather than a first-order requirement.

7 What Next?

It was agreed we should build on the use cases in Charles' discussion paper in order to illustrate the requirements and advantages of using identifiers. Phil agreed to represent some of these diagrammatically in the way which we had doing during the meeting. This work should feed into wider discussions on identifiers.

We need to make some form of recommendation to X4L projects, LTSN/RDN partnerships etc. It was felt that we needed something more concrete than advice in the form of a list of requirements and suggested scheme which might meet these requirements. We were did not wish to mandate DOIs as the preferred identifier before the costs and benefits of using them were clear, and yet the timescales of these projects do not seem to allow much time for this. In this context a paper by Andy Powell on PURL-based Object Identifiers [8] was discussed. POIs use the Open Archives Inititiative[9] metadata identifier format, which would be in line with the use of the OAI Protocol for Metadata Harvesting in the LTSN/RDN partnerships. It was agreed that we would consult X4L, the LTSN and RDN (and the wider community involved in the Jisc Information Environment) on the use of POIs for X4L and LTSN/RDN partnerships.

Time factors

We should report on the outcomes of this meeting at the next SIG meeting, which is pencilled in for the eng of August.

The next version of JORUM is due in September.

X4L will need to know about identifiers before October.

Neil McLean will be here in October to discuss identifiers in an IMS context.

[1] X4L: http://www.jisc.ac.uk/index.cfm?name=programme_x4l

[2] RDN: http://www.rdn.ac.uk LTSN: http://www.ltsn.ac.uk. Both describe learning resources such as websites, the interoperability projects are aimed at sharing these descriptions. There is the added complication that the cataloguers of the RDN or LTSN do not own the resource they are describing and so cannot assign it a definitive identifier. They do however own the metadata record.

[3] Learning and Teaching Portal: http://www.ltsn.ac.uk/genericcentre/index.asp?id=18262&

[4] UK Common Metadata Profile: http://www.cetis.ac.uk/profiles/ukcmf/

[5] Handles: http://www.handle.net/

[6] IDF: http://www.doi.org/welcome.html

[7] IDD: http://www.indecs.org/pdf/data_dict.pdf

[8] POIs: http://www.ukoln.ac.uk/distributed-systems/poi/

[9] OAI: http://www.openarchives.org/