Attatchment resource_schemas+templates030613.tar contains:

resourceRegistry.xsd
identity.xsd and .xml (template)
curation.xsd and .xml (template)
content.xsd and .xml (template)
service.xsd and .xml (template)

MORE COMMENTS TO COME

Not fully specified/restricted but they work. Outstanding questions include:

• In the schema:
  • How to use ID/IDREF (e.g. in content.xsd)
  • How to specify the root in any .xsd
  • Are there better fixes for the times when e.g. type="xs:float" insists on minOccurs >0 (have to use NaN or (-)INF) and when type="xs:string" ignors minOccurs 1 (have to use a minLength)
  • Is there a better way of making sure a polygon has min. 3 angles?
  • After much huffing I have omitted the 'associated data sets' elements for now, will consider re-instating once I have a clearer idea of requirements.
• General:
  • We need to have namespaces(?) for UCDs and Keywords
  • Units - see RegistryUnits - suggestions for initial restricted selection, based on initial catalogues. Need feeback!
    • Is there a cframe appropriate for solar data?
    • At present, I have treated spectralBandpass e.g. U, B, V as labels which do not have to be interpreted, but are available for 'dumb' matching. To enable seaching by wavelength more exactly than optical, infrared etc., we will use coverageSpectralCentral/Minimum/MaximumWavelength ?
    • I have converted everything to do with sensitivity (crudely) to Jy
    • In some cases e.g. spectral resolution, dimensionless fractions make more sense than units (see examples for USNOB etc.)
  • Conventions for words in restricted/searchable lists:
    • I suggest that we use British spellings e.g. colour, sulphur and make sure any scripts recognise and translate color etc.
    • Capitalisation - see suggestions from Elizabeth RegistryIt02Metadata at the top.
• Access to data
  • In identity and service schemas, various URLs/URIs are called for. These should include the location AstroGrid will use to access the data, if direct access exists. Please can someone supply!
• Suggested next steps - these schema are based on Hanisch V0.7 with very few additions. The actual catalogues and science cases I have looked at suggest these additional elements/changes would be useful:
  • startTime etc. is usually actually a date; specifying it as a time implies too much precision
  • For archives of pointed observations/observational data taken in multiple separate wavebands/periods, there should be spatial, spectral and temporal fractionalCoverage (within the limits of Box etc., max/min wavelength, start/stop time)
  • regionOfRegard should be expanded to spatial, spectral and temporal search region sizes.

-- AnitaRichards - 13 Jun 2003