CSV-W Profile and Triple-Store: Graphs Graphs Graphs

A real short note on where stuff goes

The example CSVW I've provided defines:

<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time> which is the dcat:Dataset
<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time.csv> as a distribution
<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/datacube> as a distribution The analogy I have is, if a data catalogue is a filing cabinet, the CSVW is a file within it. You should be able to take that file out and place it into a different filing cabinet. So the CSVW doesn't mention catalogue records or catalogues - it just describes the dcat:Dataset (and its distributions). So with that in mind I think a CSVW should only describe a csvw:Table and so we don't make use of csvw:TableGroup. To get this into PMD I'd propose:
1. The triples within the CSVW, are interpreted as JSON-LD and added to a graph within PMD with the name <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/record>.
2. The PMDCAT triples are constructed and added to the same graph.
```
<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/record> {

<http://gss-data.org.uk/catalog/datasets> dcat:record <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/record> .

<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/record> a dcat:CatalogRecord ;
    foaf:primaryTopic <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time> ;
    pmdcat:metadataGraph <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/record> .

<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time> a pmdcat:Dataset ;
    pmdcat:datasetContents <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/datacube> ;
    pmdcat:graph <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/datacube> .

<http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/datacube> a pmdcat:DataCube .

}
```
3. The observations (which are the result of csv2rdf in minimal mode) are added to a graph within PMD with the name <http://data.gov.uk/dataset/life-expectancy-by-region-sex-and-time/datacube>.
4. You could imagine for concept schemes a similar idea. There's a question of whether a skos:ConceptScheme is a distribution of a dcat:Dataset but I think it is. So,
The dcat:Catalog is handled by PMD.
We still construct the catalogue record (for now) but this remains outside of the CSVW.
The graphs reuse IRIs of the catalogue record and datacube. Also,
Versioning is handled by PMD, outside of the CSVW.
Anything requiring some compute (like a checksum) is handled outside of the CSVW.

ONSdigital / application-profile

CSV-W Profile and Triple-Store: Graphs Graphs Graphs #22

A real short note on where stuff goes