« UK Federation reaches 100 | Main | Facebook campaign helps force bank u-turn »

August 29, 2007

Conference Proper, Day One (DC-2007 Third Installment)

I'd intended to have a go at "live blogging" the first day proper of the DC-2007 conference yesterday, but the combination of my inability to produce rapid notes remotely worthy of exposing to a wider readership and a somewhat intermittent wireless connection (ah, yes, the old chestnut of blaming the technology) means that I ended up writing this at the end of the day instead.

The opening keynote was by Johannes Keizer of the Food and Agriculture Organization of the United Nations. Johannes presented a historical account of how metadata, and particularly metadata based on Dublin Core, had enabled the FAO to "collect, analyse, interpret and make available information" in support of its goal of combatting world hunger. He traced a path in which the FAO had moved from a focus primarily on describing "document-like objects" (through the AGRIS DC application profile) through the development of the AGROVOC subject thesaurus, towards a more formal ontology-based approach involving a range of resource types and their relationships. He argued that the semantic web should/would be driven by such requirements to integrate existing data-oriented applications, and that, in the FAO case, while it remained difficult to quantify, he felt that the ability to produce smarter applications based on the richer data now available was offering a significant return on the investment in the development of metadata schemas and ontologies.

The first paper session featured two papers, the first, by Akira Miyazawa (NII, Japan and member of the DCMI Usage Board) arguing persuasively that the widespread use of parallel writing systems, particularly in Asian cultures, made a case for considering objects such as "titles" not as simple character strings (literals) but as "abstractions" each of which may be associated with multiple different character strings. (Akira's argument provides an interesting perspective on the current proposal to assign a range of literal to the DCMI property dcterms:title.)

The second presentation, by Fredrik Ennokson (Royal Institute of Technology, Sweden), gave a summary of work done by Frdrik and his colleagues within the LUISA project on the development of "Annotation Profiles" as a mechanism for supporting the configuration of editing tools for RDF data. By combining a specification of patterns in the RDF graph on the one hand and a specification of the user interface (a form template) on the other, forms-based tools can be easily reconfigured to offer different interfaces - including the option of providing different interfaces to the same data reflecting the different roles/access permissions of different individuals maintaining parts of that data.

In the opening session of the afternoon, Mikael Nilsson (Royal Institute of Technology, Sweden) introduced his work on the specification for a Description Set Profile as "a way of describing structural constraints on a description set. It constrains the resources that may be described by descriptions in the description set, the properties that may be used, and the ways a value surrogate may be given."

A DSP is composed of a hierarchically organised set of "templates", mirroring the structure of the DC description set defined by the description model of the DCMI Abstract Model. So a DSP consists of a number of "Description Templates", made up in turn of "Statement Templates", which themselves contain a set of constraints either on a "Literal Value Surrogate" or on a "Non-Literal Value Surrogate". Mikael gave what I thought was a very helpful graphical presentation of the "templating" approach, starting from the full set of constructs supported by the DCAM description model, and removing or constraining the use of individual components as required by the application context. He also - a live demo! Brave man! ;-) - illustrated how a DSP could be used to dynamically configure a metadata authoring tool.

Mikael emphasised that the DSP is a formalisation of, rather than a replacement for, the "traditional" notion of what DCMI has called a DC application profile, as described, for example by the Dublin Core Application Profile Guidelines developed by CEN MMI-DC. Or perhaps more accurately the DSP reflects one core part of what constitutes a DCAP: Mikael also set the notion of the DSP within the context of the broader framework which I referred to in my previous post, where the DSP is just one component within a larger package of inter-dependent components.

Fredrik followed on from Mikael with a description of a text syntax for representing a DSP, for which he has developed a MoinMoin extension, with the result that the syntax can be used in a MoinMoin Wiki, and the Wiki can generate both a tabular HTML view and an XML representation of the data. Finally, I gave a very quick summary of work done by Julie Allinson (formerly of UKOLN, now of the University of York - who unfortunately wasn't able to attend the conference) in applying the DSP model and the DSP Wiki syntax to the case of the ePrints DCAP. I tried to emphasise that the DSP model was a mechanism for making explicit information that was already provided in the "traditional" DC application profile - as long as such a profile is firmly based on the DCMI Abstract Model. My slides are available from Slideshare.


TrackBack URL for this entry:

Listed below are links to weblogs that reference Conference Proper, Day One (DC-2007 Third Installment):


The comments to this entry are closed.



eFoundations is powered by TypePad