Self-description of non-NeXus terms (SECoP)

[g-guenther]

As a result of the SECoP mapping to NeXus the need for a standardized annotation for the meaning of a (data) field arises, e. g. using a term of an ontology. This belongs to fields that are part of the base class definition (e. g. NXsensor/measurement) as well as to those not defined by base classes. It could be useful in general to make use of the sophisticated knowledge representations that exist. To improve machine-readability of a non-NeXus field, the following information could be provided :

• data_type attribute: defines the NXDL data type of the field
• identifier attribute: A PID of a resource defining the meaning of the field
• units_identifier attribute: A PID of a resource defining the units of the field (could be provided in addition to the units attribute)

As it is somehow related, the example of #1335 is used:

entry:NXentry
    ...
    instrument:NXinstrument
        ...
        neutron_guide:NXguide
            ...
            pressure_valve_1=0.1
                @data_type="NX_FLOAT"
                @identifier="https://qudt.org/vocab/quantitykind/Pressure"
                @units="mbar"
                @units_identifier="http://qudt.org/vocab/unit/MilliBAR"

I noticed similar demands in other projects and I think it could be of general interest. Would that make sense?

[g-guenther]

For my understanding and to elaborate this topic (see also PR #1570), I present some examples which could help to find a proper implementation.

Draft 1: Current NXobject definition

The current definition of NXobject which allows to annotate NeXus base classes (HDF groups).

entry:NXentry
    ...
    sample:NXsample
        ...
        identifier_sample="http://nmrML.org/nmrCV#NMR:1000457"
            @type="ARK"
        humidity=0.3
            @units="%RH"
        humidity_env:NXenvironment
            ...
            identifier_humidity_env="http://purl.obolibrary.org/obo/ENVO_01000254"
                @type="ARK"
            humidity_sensor_1:NXsensor
                ...
                identifier_humidity_sensor_1="http://purl.obolibrary.org/obo/CHMO_0002793"
                    @type="ARK"
                measurement="humidity"
                    @custom=True
                value=0.3
                    @units="%RH"

Draft 2: Current NXobject definition extended

Same example as Draft 1 but the current definition of NXobject is extended to fields and attributes.
This can't be realized in a consistent way since there are no attributes for attributes. In this example,
the 'identifierNAME' construct for the 'units' attribute is flattened using the attributes 'identifier_units'
and 'identifier_type_units'.

entry:NXentry
    ...
    sample:NXsample
        ...
        identifier_sample="http://nmrML.org/nmrCV#NMR:1000457"
            @type="ARK"
        humidity=0.3
            @units="%RH"
            @identifier_units="http://qudt.org/vocab/unit/PERCENT_RH"
            @identifier_type_units="ARK"
        identifier_humidity="https://qudt.org/vocab/quantitykind/RelativeHumidity"
            @type="ARK"
        humidity_env:NXenvironment
            ...
            identifier_humidity_env="http://purl.obolibrary.org/obo/ENVO_01000254"
                @type="ARK"
            humidity_sensor_1:NXsensor
                ...
                identifier_humidity_sensor_1="http://purl.obolibrary.org/obo/CHMO_0002793"
                    @type="ARK"
                measurement="humidity"
                    @custom=True
                identifier_measurement="https://qudt.org/vocab/quantitykind/RelativeHumidity"
                    @type="ARK"
                value=0.3
                    @units="%RH"
                    @identifier_units="http://qudt.org/vocab/unit/PERCENT_RH"
                    @identifier_type_units="ARK"
                identifier_value="https://qudt.org/vocab/quantitykind/RelativeHumidity"
                    @type="ARK"

[g-guenther]

Here two questions arise:

• Question 1: The NXobject definition contains a field 'identifierNAME' which is interpreted here as
  'identifier_GROUPNAME'. Is this correct?
• Question2: Does the 'identifierNAME' field contain the complete, web-resolvable address, e.g. in case of an ORCID

entry:NXentry
    ...
    user:NXuser
        ...
        identifier_user="https://orcid.org/0000-0002-1825-0097"
            @type="ORCID"

, or does the field only contain the identifier e.g.

entry:NXentry
    ...
    user:NXuser
        ...
        identifier_user="0000-0002-1825-0097"
            @type="ORCID"

?

[g-guenther]

To record a comment from the meeting on July 16:

• data_type attribute would be redundant since the data type is accessible from HDF5

[g-guenther]

Draft 3: Attributes

Same example as Draft 2 but the identifiers are implemented as attributes (if possible).
This can't be realized in a consistent way, too, since there are no attributes for attributes in HDF5.
For groups (base classes) and fields the attributes could be:

• identifier
• identifier_type

For attributes (ATTRIBUTENAME), the attributes could follow the schema:

• ATTRIBUTENAME_identifier
• ATTRIBUTENAME_identifier_type

Advantage:

• follows the NeXus data model more closely with identifiers closely attached to the subject they describe

Disadvantage:

• only one identifier for a group, field, and attribute

entry:NXentry
    ...
    sample:NXsample
        @identifier="http://nmrML.org/nmrCV#NMR:1000457"
        @identifier_type="ARK"
        ...
        humidity=0.3
                @identifier="https://qudt.org/vocab/quantitykind/RelativeHumidity"
                @identifier_type="ARK"
                @units="%RH"
                @units_identifier="http://qudt.org/vocab/unit/PERCENT_RH"
                @units_identifier_type="ARK"
        humidity_env:NXenvironment
            @identifier="http://purl.obolibrary.org/obo/ENVO_01000254"
            @identifier_type="ARK"
            ...
            humidity_sensor_1:NXsensor
                @identifier="http://purl.obolibrary.org/obo/CHMO_0002793"
                @identifier_type="ARK"
                ...
                measurement="humidity"
                    @custom=True
                    @identifier="https://qudt.org/vocab/quantitykind/RelativeHumidity"
                    @identifier_type="ARK"
                value=0.3
                    @identifier="https://qudt.org/vocab/quantitykind/RelativeHumidity"
                    @identifier_type="ARK"
                    @units="%RH"
                    @units_identifier="http://qudt.org/vocab/unit/PERCENT_RH"
                    @units_identifier_type="ARK"