This application profile seems to assume that it's running on a server where data can be partitioned into graphs, connected to catalogues and then the dcat:Dataset records inside those catalogues can be retrospectively updated to declare that they are referenced somewhere externally (dcterms:isReferencedBy).
It goes on to partially discuss how some of this data should be held inside CSV-W files (talks a fair bit about use of columns/virtual columns to represent units, etc.).
Essentially, a lot of the content of this document seems to define how an RDF-focused application server should represent the data that has been uploaded to it in the form of CSV-Ws and also tries to define how the data should be held inside a CSV-W - but it doesn't make at all clear which triples belong in the initial CSV-W and which triples are maintained/updated by the application server.
I think we need:
A single document describing how CSV-Ws are structured, detailing the triples to be contained there-in
A separate document describing how the data should be structured on an RDF-aware application server that's able to do creative things like deal propertly with versioning and linking the data sets into the catalogues relevant to that server.
This application profile seems to assume that it's running on a server where data can be partitioned into graphs, connected to catalogues and then the
dcat:Datasetrecords inside those catalogues can be retrospectively updated to declare that they are referenced somewhere externally (dcterms:isReferencedBy).It goes on to partially discuss how some of this data should be held inside CSV-W files (talks a fair bit about use of columns/virtual columns to represent units, etc.).
Essentially, a lot of the content of this document seems to define how an RDF-focused application server should represent the data that has been uploaded to it in the form of CSV-Ws and also tries to define how the data should be held inside a CSV-W - but it doesn't make at all clear which triples belong in the initial CSV-W and which triples are maintained/updated by the application server.
I think we need: