Linking research & learning technologies through standards

Sharing learning content between schools, jurisdictions, sectors—even countries—is an efficiency that makes sense, and is coming to make more and more sense, as there is more content to share, and more people needing content, with less time to write it. Content can be discovered on the open web, through Google, but that kind of discovery is not particularly targeted, quality-controlled, or inclusive. Learning content is normally discovered through repositories, with well-ordered, authoritative, and searchable metadata, rankings and ratings; with established authentication, authorisation and licensing; and all the value adding a controlled environment allows.

That’s all very well if you’re happy to stick to what’s locally available. If you’re not, you need to find out what relevant content there might be in other repositories, which you don’t yet have access to. In fact, even before that, you need to find out which other repositories there is any point in looking for at all. Repository federation, as seen in LORN in the vocational sector, deals with the problem by bringing the other repositories to you (and vice versa).

But there is no one federation of learning content repositories spanning the globe, and there is always something out there that could end up useful; so the problem of finding content elsewhere gets pushed up a level. Even within a federation, the repositories are still autonomously updating and enhancing content, and still need infrastructure to synchronise with each other, and exchange content. And by the time the federations turn into multinational federations of federations, as with the Learning Resource Exchange from European Schoolnet, or the GLOBE alliance, making content exchange scale over large distances and numbers becomes a pressing priority.

Τhe IMS Learning Object Discovery and Exchange (LODE) project group has been set up to help deal with this kind of issue. It intends to create consistent descriptions of registries of learning content, and services to discover and access such content. Crucially, it concentrates not only on ways of better discovering content through registries, but also on ways of discovering those registries themselves. This depends critically on metadata describing content registries—what kind of content they have, what fields they cover, how frequently they are updated, how their content is licensed. The metadata required to drive that kind of discovery is not a million miles away from how libraries and registries are described in general, and elsewhere we describe how we profiled the ISO 2146 standard for registry services to describe learning content registries under LODE.

• Note: There is an extensive post elsewhere on how European Schoolnet is harnessing the ASPECT project and IMS LODE to meet its goals of making learning content discoverable across the European Union. Link Affiliates are co-chairs of the LODE activity, and have spent a lot of time on repository federation infrastructure as something indispensable to improving education outcomes. (Hence for example the Federated Repositories in Education (FRED), project undertaken in 2007.) An Novermber 2008 summary of LODE work and Link Affiliates’ involvement is also available.

As part of our engagement with LODE, we are testing the CQL Context Set for SRU-based resource search, developed by LODE. We are also doing proof of concept development on using the FRBR model to cluster together search results for different versions, formats and copies of the same learning content. The development work is continuing apace at the USQ Advanced Digital Futures Institute, and the development wiki is publicly accessible.

As in other sectors, schools have long been burdened with the incompatibility of the multiple IT systems used to run their business. The Learning Management Systems, the Student Enrolment Systems, the systems dealing with assessment, pastoral care, attendance, staffing, timetabling—all of these store data about the same students in different ways, and each exports data in its own way. To get the systems to share data between each other has often meant costly custom porting for each pair—where it has not involved printing the data out, and rekeying it from scratch. Waving printouts at a keyboard is not, of course, fulfilling the promise of the paperless office, and it hardly translates to data at one’s fingertips.

The school sector realised quite early (1998) that something could be done about this, and the Systems Interoperability Framework (SIF) was developed Stateside in response to it. SIF was developed several years before Service-Oriented Architecture started to address similar issues of data incompatibility in industry, but it takes an approach similar enough that it can be stated in SOA terms. Data is exchanged between systems across a common trust environment, using common data structures in XML—just like the Enterprise Service Bus and SOAP of SOA. Systems are able to exchange data because agents translate their native data to the common formats and back again. Data can be pulled in, in a request–response pattern, or pushed out, in a subscription pattern. Unlike SOA, the protocols and common data models are standardised and fixed ahead of time for the school domain, and do not require the systems to be reengineered to fit the system protocols better; so SIF can be layered over existing systems relatively straightforwardly.

The recent trend in the Australian government school sector, and to a growing extent in the Catholic sector, has been to host school systems centrally; this leads to greater efficiencies and security in how data is handled and exchanged, and relieves schools from the burden of having to run systems themselves. However, jurisdictions still have to deal with multiple systems internally, some more centralised than others. They also have occasion to exchange data with other jurisdictions and schools, especially when students move interstate, or in dealing with national testing and benchmarking. Dealing with these issues has made SIF an attractive proposition for the Australian school sector, whether to support the integration of all their internal systems, as is taking place in Victoria, or to provide a consistent outward interface to the data they are authorised to share. This has led to the SIF Association AU initiative, led by representatives from all Australian school systems.

As a relatively lightweight technical architecture, SIF does not particularly depend on where it is deployed, and there have already been several successful deployments of SIF in the UK, in addition to the pilots now underway in Australia. What does need to change from place to place is the model underlying the common data format that the system uses. The original SIF data model deals with the realities of the American school system, so it represents data that makes sense for that context. Because students are provided lunch at school, the logistics of school canteens are a major concern of SIF data modelling, and there are obvious dollar and cent efficiencies in getting the canteen system to talk to the student attendance system. When Australian school systems exchange information, canteen logistics are not a major concern; but getting timetabling right is.

Likewise, the data fields and values captured in SIF reflect American conventions and requirements: they deal in quarters and quinmesters, charter schools, and demographics driven by the American Census and the NCES. The data collected and exchanged in Australian schools needs to reflect Australian requirements, and to conform to Australian norms and conventions. So data is about ESL, not English Proficiency, and the codes for countries and languages are the Australian Bureau of Statistics’. Dealing with these codes in turn means addressing questions such as how many different shades of “Not Applicable” to allow for Yes/No fields, or whether to include both the nodes and the leaves in a hierarchical vocabulary (e.g. whether to allow Netherlandic as a language choice, or only its child nodes, Dutch and Frisian).

Link Affiliates and the ABS, along with members of the SIF-AU data working group, gave feedback on how best to specify the data objects and vocabularies to be used. The SIF-AU draft standard is now being vetted by SIF; meanwhile, pilot projects are underway in several jurisdictions, with an aim to finalise work by the end of the year.

The ISO 2146 standard defines a conceptual framework for registry services. It lays out a model of parties, collections, services, and activities, and how they all relate to each other, to capture the process of a library or repository going about its business of collecting content, and making content available. The standard originated in the library sphere, but is being written in a generic enough fashion to apply across a broad range of registries, both physical and digital, and across a range of disciplines. At the same time, it is intelligently specific in the attributes and relations it defines, so it allows reasonably detailed descriptions of what is going on in a registry. These descriptions can be the basis for cogent and useable metadata on collections in general, and on how they are presented to the world.

The new draft of ISO 2146 is still in progress (with the core participation of Link Affiliate Judith Pearce). But because of its advantages, it has already seen use in several domains. The Online Research Collections Australia (ORCA), under the Australian Partnership for Sustainable Reposiories (APSR), has used ISO 2146 as the basis for its descriptions of research collections in Australia, to be used for better discovery of both collections and services. ORCA, in turn, will form a major part of the Australian National Data Service (ANDS)’ Data Commons, a space where researchers nationally can share, discover, and build on the available collections of each others’ work.

In the e-learning space, the IMS Learning Object Discovery and Exchange project group has been set up to create consistent descriptions of registries of learning content, and services to discover and access such content. The collection descriptions that will be accessed through LODE all involve learning objects, but these can be of very different sorts, and with very heterogeneous content; IMS LODE is looking to be applied in contexts like the ASPECT project, or the Globe alliance, coordinating registries of learning content at various levels of schooling, from many countries.

Link Affiliates participates in IMS LODE. As the conceptual model for LODE description of registries was being drawn up, we believed that a profile of ISO 2146, customised to the particular requirements of e-learning registries, would be a suitable basis for it. Customising the ISO 2146 is consistent with how the scheme is meant to be used: it is a framework for making sense of the registry world, rather than a uniform schema—which is why it does not define an XML serialisation. That said, IMS LODE ISO 2146 is a rather elaborate model: it defines well over forty entities to describe the many ways in which registry entities can relate to each other, with complex relations between them. Cutting them down to a profile applicable to our context needed us to get a quick overview of what the model was expressing, and what the consequences were of eliminating or adding entities.

Read the rest of this entry »