Metacat

   * `oboe:Measurement <http://ecoinformatics.org/oboe/oboe.1.0/oboe-core.owl#Measurement>`_ : The primary body of semantic annotations on attributes.

   * `oboe:ofCharacteristic <http://ecoinformatics.org/oboe/oboe.1.0/oboe-core.owl#ofCharacteristic>`_ : Specifies which Characteristic (subclass) is measured

   * `oboe:usesStandard <http://ecoinformatics.org/oboe/oboe.1.0/oboe-core.owl#usesStandard>`_ : Specifies in which Standard (Unit subclass) the measurement is recorded.

Model details

--------------

Using the ``weight`` column in our example data package, we can illustrate the annotation model's use of concepts from OBOE, OA, and PROV.

The primary entry point for the annotation is ``#ann.4.1`` and was asserted by Ben Leinfelder (foaf:name) , identified with his ORCID URI (oa:annotatedBy).

The body of the annotation (oa:hasBody) is comprised of an oboe:Measurement instance, ``#weight``, that measures ``Mass`` (oboe:ofCharacteristic) in ``Gram`` (oboe:usesStandard).

The target of the annotation (oa:hasTarget) points to the EML metadata resource (oa:hasSource) that documents the data table and selects a particular part of the metadata that describes 

the specific ``weight`` data attribute (oa:hasSelector). Because the EML metadata is serialized as XML, we can use an XPath oa:FragmentSelector to identify the data column being annotated.

Note that our XPath expression identifies ``weight`` as the second column in the first data table in the data package: #xpointer(/eml/dataSet/dataTable[1]/attributeList/attribute[2].

In order to bind the column annotation of the metadata to the physical data object (the three-column CSV file), we need to traverse the packaging model where an additional annotation expresses the relationship 

between the data and metadata objects. The annotation, ``#ann.1.1``, asserts that the Metadata file (#eml.1.1) describes (cito:documents) the data file (#data.1.1). More specifically, the annotation target specifies 

where in the EML the #data.1.1 object is described by using an oa:FragmentSelector with an XPath pointer to the first data file documented in the EML: #xpointer(/eml/dataSet/dataTable[1].

Note that the annotation model uses a slightly different model than the original ORE resource map model recommended by DataONE. While it is more complicated to include pointers to data documentation within the metadata,

we have found that the current ORE maps are not sufficiently descriptive on their own and any consumers must also consult the metadata to figure out which object is the csv, which is the pdf, which is the script, etc...

By incorporating the metadata pointer within the annotation model, we hope to be able to hanlde data packages that use manu different metadata serializations without having to write custom handlers for each formatId.

The Metacat Index component has been enhanced to parse semantic models provided as RDF. 

The general purpose RdfXmlSubprocessor can be used with SparqlFields to extract key concepts from any given model that is added to the Metacat MN document store.

The processor assumes that the identifier of the RDF document is the name of the graph being inserted into the triple store and provides that graph name to the query engine for substitution in any query syntax ($GRAPH_NAME).

Index Fields 

_________________

Currently, these dynamic, multi-valued string fields allow us to index the new semantic content without changing the SOLR schema. 

They are multi-valued because they will store the entire class subsumption hierarchy (up) for any matching concepts

and because they will store annotations from the same metadata resources for different attributes.

	* ``characteristic_sm`` - indexes the oboe:Characteristic[s] for oboe:Measurement[s] in the datapackage

	* ``standard_sm`` - indexes the oboe:Standard[s] for oboe:Measurement[s] in the datapackage

Continuing with example model, these concepts would be indexed for the data attributes described in the datapackage metadata.

_______