Test schema org

[jyucsiro]

First cut.

See #80

[jyucsiro]

Allowing for netCDF metadata on the web to be tested, e.g. URLs passed in like so:

$ python nc2rdf.py -o json-ld --schema-org https://www.ngdc.noaa.gov/thredds/dodsC/arctic/Polar-APP-X_v01r01_Nhem_1400_d20160801_c20160803.nc

result:

{
    "@context": "http://schema.org/",
    "description": "The Extended AVHRR Polar Pathfinder (APP-x) version-2 Thematic Climate Data Record (CDR) includes surface temperature, surface albedo, surface and the Top Of the Atmosphere (TOA) shortwave and longwave radiative fluxes, cloud properties (amount, phase, particle size, optical depth,top pressure and temperature, surface and TOA radiative effect), and ice thickness and age. The APP-x CDR has twice daily data at local solar times of 14 and 04(02) for the Arctic(Antarctic) at a spatial resolution of 25 km for both poles.",
    "http://schema.org/identifier": "Polar-APP-X_v01r01_Nhem_1400_d20160801_c20160803.nc",
    "http://schema.org/license": "No restrictions on access or use",
    "id": "_:N1ffe69b6dd6c4ceb895f7c79d3c17b2d",
    "keywords": "EARTH SCIENCE > ATMOSPHERE > ATMOSPHERIC RADIATION > SOLAR RADIATION, EARTH SCIENCE > ATMOSPHERE > ATMOSPHERIC RADIATION > LONGWAVE RADIATION, EARTH SCIENCE > ATMOSPHERE > ATMOSPHERIC RADIATION > RADIATIVE FLUX, EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD MICROPHYSICS > CLOUD DROPLET CONCENTRATION(SIZE), EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD MICROPHYSICS > CLOUD OPTICAL DEPTH(THICKNESS), EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD MICROPHYSICS > CLOUD DROPLET PHASE, EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD PROPERTIES > CLOUD TOP PRESSURE, EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD PROPERTIES > CLOUD TOP TEMPERATURE, EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD PROPERTIES > CLOUD TYPE, EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD PROPERTIES > CLOUD FRACTION, EARTH SCIENCE > ATMOSPHERE > CLOUDS > CLOUD RADIATIVE TRANSFER > CLOUD RADIATIVE FORCING, EARTH SCIENCE > CLIMATE INDICATORS > CRYOSPHERIC INDICATORS > ICE DEPTH(THICKNESS), EARTH SCIENCE > CRYOSPHERE > SEA ICE > ICE DEPTH(THICKNESS),EARTH SCIENCE > CRYOSPHERE > SNOW and ICE > ALBEDO, EARTH SCIENCE > CRYOSPHERE > SNOW and ICE > SNOW and ICE TEMPERATURE, EARTH SCIENCE > LAND SURFACE > LAND TEMPERATURE, EARTH SCIENCE > LAND SURFACE > LAND ALBEDO, EARTH SCIENCE > OCEANS > SEA ICE > ICE DEPTH(THICKNESS),EARTH SCIENCE > TERRESTRIAL HYDROSPHERE > SNOW and ICE > ALBEDO, EARTH SCIENCE > TERRESTRIAL HYDROSPHERE > SNOW and ICE > ICE DEPTH(THICKNESS), ",
    "name": "Extended AVHRR Polar Pathfinder Fundamental Climate Data Record (APPx CDR)",
    "type": "Dataset"
}

[adamml]

@jyucsiro I think you should (strictly) preface the "type" and "id" keys with @ symbols, such that they become "@type" and "@id". Without, they parse ok but with they are reserved in JSON-LD for the explicit definitions.

On the identifier and license and keys, the "http://schema.org/" is not required.

As we discussed on the call, it would be nice to split those GCMD keywords into an array...

[marqh]

do both the hdf and netcdf APIs support loading from URI?

if so, we should just let them do that

is that what you are doing in this case? does it already work?

[jyucsiro]

I haven't checked with HDF, but it seems to work with the netCDF APIs.

Yes, that's what I'm doing in this case - both local and from URLs (mainly netCDF files served through thredds).

[jyucsiro]

@jyucsiro I think you should (strictly) preface the "type" and "id" keys with @ symbols, such that they become "@type" and "@id". Without, they parse ok but with they are reserved in JSON-LD for the explicit definitions.

Thanks @adamml. I'm using the rdflib-json-ld serializer (https://github.com/RDFLib/rdflib-jsonld). For some reason, for the example above, it doesn't add "type" and "id" with those symbols. Tinkering around https://json-ld.org/playground/, it looks like the output is a 'compacted' version, where "type" and "id" are implicitly equivalent to "@type" and "@id"?

I tried getting the json-ld serializer to output with those '@' symbols but am not getting much success.

On the identifier and license and keys, the "http://schema.org/" is not required.

Can you clarify? I'm not sure where to find this.
I see what you mean - it was from another example on #80

The issue will be then which namespace the identifier/license properties will come from. This link (https://developers.google.com/search/docs/data-types/dataset) lists license and identifier...

As we discussed on the call, it would be nice to split those GCMD keywords into an array...

Yes, that would be good.

It would need to be either at the time it gets parsed into the native BALD graph or in converting to schema.org?

I've implemented this in nc2rdf for now. @marqh what do you think?

[adamml]

@jyucsiro

The issue will be then which namespace the identifier/license properties will come from. This link (https://developers.google.com/search/docs/data-types/dataset) lists license and identifier...

Because you have {"@context": "http://schema.org",...} without the identifier/license properties, you are saying use the schema.org namespace.

[marqh]

thank you for all the input, this looks good to go