CellML Metadata

The CellML metadata specification and CellML API

This is something we need to debate in the workshop. My own personal opinion is as follows:

  • we need a process to specify recommendations for community wide metadata to stop arbitrary RDF schemas being designed where the intention or current specification in CellML metadata already provides for it or has some base to work from.
  • we should write down all the proposals we think need to be addressed - for example: workflow states, simulation, graphing, experimental data, etc.
  • we should make sure the correct way to represent community wide metadata is well specified through RDF schemas and perhaps other data modeling formats such as UML.
  • we need to make sure there is a method to produce a canonical XML form of metadata so that XSLT/XPATH methods are still useable in a sane way - I imagine something like taking an accepted rdf library, producing rdf-abbreviated form and then reverse engineering a DTD, w3c-XML-Schema, RelaxNG Schema and advertising this would go a long way to aiding metadata tool development for those that don't need ontologically advanced query interfaces. Most of Tommy's work could operate at that level if it needed to for instance.
  • I would prefer that models submitted to the repository that have extended metadata are also accompanied by a submission of an associated RDF Schemas so that we can see explicitly the extensions as additional assertions on existing CellML metadata classes and properties (including ones not in our namespace) and begin a dialogue with them about it.
  • the CellML API should offer at least two kinds of metadata interface:
    1. RDF aware - mostly about extracting RDF from a particular CellML/XML document and presenting it in one block, perhaps applying filters first.
    2. Frame based - i.e. aware of the Rdf Schemas specified and able to produce a generic object oriented frame interface to the metadata in a CellML model.