[These posts should be regarded as drafts for comment until I remove this note]

In the UKOER 2 programme Strand C funded “Projects identifying, collecting and promoting collections of OER and other material around a common theme” with the aim “…to investigate how thematic and subject area presentation of OER material can make resources more discoverable by those working in these areas” (UKOER 2 call document). The projects had to create what were termed static and dynamic collections of OER. The intent of the static collection was that it could in some way act as an identity, focus, or seed for the dynamic collection. Six projects were funded: CSAP OER, Oerbital, DelOREs, Triton, EALCFO, Open Fieldwork and a range of approaches and technologies was taken to making both static and dynamic collections. The projects are all worth reading about in more detail – however, in this context there are two possible general patterns worth considering.

Technology

Overview of technical choices in UKOER 2 Strand C

The above graph shows the range of technology used in the Strand. Although a lot could (and should) be said about each project individually when their choices are viewed in aggregate the following technologies are seeing the widest use.

Graph of technologies and standards in us by 50% or more of Strand C projects

Although aspects of the call might have shaped the projects’ technical choices to some extent, a few things stand out:

• the focus on RSS/Atom feeds and tools to manipulate them
  • reflection: this matches the approach taken by many of the other  aggregators and discovery services  for OER and other learning materials as well as the built in capabilities of a number of the platforms in use [nb “syndicated via RSS/Atom” was a programme requirement]
• a relative lack of a use of OAI-PMH
  • reflection: is this indicative of how many content providers and aggregators in the learning material’s consume or output OAI-PMH?
• substantial use or investigation of wordpress and custom databases (with php frontends)
  • reflection: are repositories irrelevant here because they don’t offer easy ways to add plugins or aggregate others’ content (or are there other factors which make WordPress and a custom database more appealing)

Community

One of the critical issues for all of these projects in the creation of these collections has been the role of community; for some of the strand projects the subject community played a crucial role in developing the static collection which then fed, framed, or seeded the dynamic collection, for other projects the subject community formed the basis of contributing resources to the dynamic collection.

Although the projects had to be “closely aligned with relevant subject or thematic networks – for example Academy Subject Centres, professional bodies and national subject associations” , I find it striking that many of the projects made those defined communities an integral part of their discovery process and not just an audience or defining domain.

Reflections on community

I’m hoping someone else is able to explore the role of community in discovery services more fully (if not I’ll try to come back to this)  but I’ve been struck by the model used by some projects in which a community platform is the hook leading to resource discovery. It’s the opposite end of the spectrum to Google – to support discovery you create a place and content accessible and relevant to a specific subject domain. The place you create both hosts new content created by a specific community and serves as a starting point to point to further resources elsewhere (whether those pointers are links, learning pathways, or tweaked plugin searches run on aggregators or repositories). This pattern mirrors any number of thriving community sites (typically?) outside of academia that happily coexist in Google’s world providing specialist sources of information and community portals  (for example about knitting, cooking, boardgames).

What it doesn’t mirror is trying to entice academics to use a repository… [I like repositories and think they’re very useful for some things , but this and the examples of layering CMSs on top of repositories, increasingly makes me think that on their own they aren’t a great point of engagement for anybody…]

[These posts should be regarded as drafts for comment until I remove this note]

Dissemination protocols

Dissemination protocols in use in the UKOER 2 programme

The chosen dissemination protocols are usually already built in the platforms in use by projects; adding or customising an RSS feed is possible but often intricate and adding an OAI-PMH feed is likely to require substantial technical development. DelOREs investigated existing OAI-PMH plugins for WordPress they could use but didn’t find anything usable within their project.

As will be discussed in more detail when considering Strand C – RSS is not only the most supported dissemination protocol, from the programme’s evidence, it is also the most used in building specialist discovery services for learning and teaching materials. The demand for an OAI-PMH interface for learning resources remains unknown. [debate!]

Jorum representation

Methods of uploading to Jorum chosen in UKOER 2 programme

• The statistics on Jorum upload method are denoted expressions of intent – projects and Jorum are still working through these options.
• Currently RSS upload to Jorum (along with all other forms of bulk upload) is set up to create a metadata record not deposit content.
• Three of the uploaders using RSS are using the edshare/eprints platform (this platform was successfully configured to deposit metadata in bulk  via RSS into Jorum in UKOER phase 1).
• Jorum uses RSS ingest as a one-time process – as I understand it it does not revisit the feed for changes or updates [TBC]
• As far as I know PORSCHE are the only project who have an arranged OAI-PMH based harvest (experimental for Jorum upload under investigation as part of an independent project – [thanks to Nick Shepherd for the update on this HEFCE-funded work: see comments and more information is available on the ACErep blog)]

• a protocol almost universally implemented in repositories and consequently (usually) publishing metadata about repository contents to the world
• a protocol frequently reviled by anyone trying to aggregate feeds from different repositories and build discovery tools and services on top of that aggregate.

I’m not going to repeat OAI-PMH’s problems in detail (PERX, the experience of the NSDL with metadata quality, Andy Powell, Jim Downing and many others have done that), suffice to say their are issues about how protocol was implemented by software, how it is used by metadata creators, and how not web-friendly it is and niche it remains.

However,  I realised recently that I’d begun to think that it must be better by now – surely the teething problems are done with – implementations more mature, record quality better, and aggregation more stable. This is, in part, because it remains the standard for sharing repository metadata and because in a number of settings it works well – there are plenty of communities using it to establish and provide services either by creating ‘closed’ controlled conditions through communally enforced ways of recording information and application profiles, guidelines and ‘political’ agreements in additional to the protocol or by creating tools that simply hack their way around whatever data they get and offer good enough services.

So it was with interest that I picked up on a discussion on twitter about what’s wrong with OAI-PMH and an upcoming paper on using Atom .[edit: I’m updating this list with fragments of conversation about OAI-PMH if I see them, and the odd link or two] It’s not the first time I’ve caught fragments of conversations on the use of feeds – for example the earlier RSS and repositories discussion.

There are a slew of issues around trying to standardise feed types (as we discovered in the discussions organised around RSS as a possible metadata deposit mechanism). See for example Feed Deposit , OER, RSS, and JorumOpen , and the two later review articles (1 , 2) as well as the email list discussions ). Given the increasing use of RSS or Atom in some of the OER discovery tools, the work listed above, and the wider promotion of it in the UKOER pilot projects, why am I so interested in this discussion about OAI-PMH and about another effort to use Atom/ RSS?

I’m happy to see this debate crop up again in the wider (library) repository community for two reasons:

1) perhaps obviously it reaffirms the issues with OAI-PMH, that they haven’t changed, and the possibilities RSS/Atom offers,

2) more importantly it’s bringing the discussion about the feeds produced by repositories into the library/ scholarly works communities. Like it or not those are the communities who are most using repositories and the communities who can to some degree shape the development of repository software and specifications. Few repository platforms natively support much customisation of the feeds they produce and until the wider repository community wants that type of functionality or control and begins to think how update or move beyond OAI-PMH* there’s little reason for repository developers to work on the problem.

Without those changes anyone wanting to manage learning materials in a repository still has to hack their own fixes, build their own repository – or not use repositories (but that’s another question).

*I should note: I like OAI-PMH – I can play with in a browser, repository explorer is a good tool, and using OAI-PMH I can get and interact with someone else’s ‘raw’ metadata. – I’m just no longer convinced it’s the right tool to share metadata – in part by how few successful discovery services there are which use it.

The Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH ; http://www.openarchives.org/OAI/openarchivesprotocol.html ) “provides an application-independent interoperability framework based on metadata harvesting.” The protocol is widely used by repository software to make metadata about the resources they store available. In its use the repository acts as a data provider which is then able to be harvested by a data harvester. Two issues of note:

1. although most repositories can function as data providers the data harvesting aspect of the protocol often requires separate software and is much less widely implemented.
2. OAI-PMH specifies a minimal base metadata set of OAI_DC (~ the simple DC element set) therefore any implementation of it should be able to provide this as a minimum. Other metadata standards such as DC Terms or IEEE LOM can also be made available for harvesting.

Although OAI-PMH is a well established standard which is widely used, at this point it’s use for open educational resources is somewhat limited. OAI-PMH is not currently in use by Jorum for metadata harvesting and, as far as I know, there are not many OAI-PMH based harvesters offering aggregated search services for educational materials (outside of those within particular closed/ or semi-closed communities). DiscoverEd from Creative Commons does offer an OAI-PMH based harvest but prefers RSS/Atom based approaches (Enhanced Search for Educational Resources – A Perspective and a Prototype from CCLearn (2009) http://learn.creativecommons.org/wp-content/uploads/2009/07/discovered-paper-17-july-2009.pdf , p12). UPDATE: Please see comment from Jenny Gray below.

OAI-PMH is being used or is supported by:

• ChemistryFM (option once content is in backup ePrints repository)
• Phorus
• Unicycle
• OCEP
• Open Exeter
• OpenStaffs
• OERP (use unknown)
• Humbox

OAI-PMH is in active use by (as opposed to out of the box support):

• Phorus (harvesting catalogued resources from Intute.)
• TRUE (using a Drupal plug in?)
• ADM OER
• ChemistryFM (option once content is in backup ePrints repository)

OAI-ORE

“Open Archives Initiative Object Reuse and Exchange (OAI-ORE) defines standards for the description and exchange of aggregations of Web resources.” (http://www.openarchives.org/ore/)

As a standard for describing aggregated or compound resources ORE has the potential to be highly relevant to some types of education materials made up of distributed web resources; its use, however, with educational materials has thus far been somewhat limited. It has however been used as an exchange mechanism for moving repository contents from one system to another.

Projects using OAI-ORE:

• ChemistryFM (export function from WordPress – will use to backup content to repository)
• ADMOER (export function from ePrints)
• HumBox (export function from ePrints)