Mar. 5, 2007
Telecon Agenda and Notes, Monday, Mar. 5, 2007
1. The AP presentation at the Usage Board meeting. Jon's questions:
- Do we have some expectation that a machine-readable AP will be used
for data validation?
- I assume yes, so... what kind of machine-readable representations we're considering -- RDF, RDFS/OWL, XSD, XML
We agreed at the outset that RDF doesn't do data validation, so that we need to do is shift somewhat towards XML from a purely RDF application. Should we settle for machine readable rather than machine usable? Stuart pointed out that the early GEM tool "assumed" an application profile and built data based on that but did not validate the data that was created. It could use elements from any number of schemas. It was also agreed that there are two problems: creating a schema that would allow a tool to produce conformant data, and allowing an application to evaluate and validate data that declares itself as following the AP. Both of these are important and within the area we want to work.
Stuart suggested that we look at the proposal for the new "DC in XML" guidelines, and determine whether we can work with the new DC encoding? In theory, anything that can be dumped into the new DCMI/XML, should be easily transportable to RDF. Jon: the question is can we register a useful AP that can be expressed in either XML or RDF. Stuart: that's what the new XML binding is intended to do, there's a schema attached to it, etc. Diane: We should present this as a test case for those guidelines, perhaps? Joe, Stuart: we need to know whether we can do this without data loss.
- Will we _require_ that an AP have subordinate metadata schemas and what form will these schemas take? I assume the form as an AP, but want to be sure
Jon: To what extent are we creating an ontology. Joe: we need to look at the way the AP says that the data is modelled. Jon: The question comes up because we may want to express the AP as RDFS/OWL, and does it bring us additional functionality for logical reasoning and data constraints.
Diane: We can use the KMODDL AP as a test case.
Discussion on the inability within an AP to discuss a domain for a property without being able to define it's context, specifically relating to a node with a URI. This is related to the sheep discussion Joe has been having on the lists. In order for a property in an AP to be defined correctly it had to be related to a URI, and there has to be a definition at the URI.
- Other questions?
- Who will present this at the UB?
- Diane has been working on the narratives, incorporated most of Jon's suggestions in the extension proposal and filled in some blanks on the new proposal. We should probably all read through them and be prepared to discuss gaps, etc.
- Diane has sent a note to Kim Lightle as suggested last week--no reply yet.
3. Stuart's report from WebWise?