CSV/JSON output of DPV v2 and extensions about dpv HOT 15 CLOSED

It would go great to have the .csv files output alongside the other exports, e.g.,

v2/
     ai/
         ai.csv
         ai.jsonld
         [...]

The rows in the CSV should be named with the same names as used elsewhere in the DPV - e.g., term,label, isSubclassOf and so on.

from dpv.

Hi. The files are in the dev branch under v1.0 and v2.0. CSV files are only provided for v2.0 https://github.com/w3c/dpv/tree/dev and the live version is at dev.dpvcg.org e.g. https://dev.dpvcg.org/v2.0/dpv/

from dpv.

@coolharsh55 thanks - I'll have a go at integrating this tomorrow. I'll leave feedback here if any.

from dpv.

For IRIs and namespaces - will the exported RDF, CSV and so on have the specific versioned IRIs - e.g., https://w3id.org/dpv/v2.0/ai#AI or the unversioned one https://w3id.org/dpv/ai#AI or both? If the definitions and/or semantics change across versions we need to track which version of a term is in use in our application.

It will contain the versioned IRIs e.g. https://www.w3id.org/dpv/v2.0/ai#AI

from dpv.

@thovden thanks for spotting that (mea culpa) - I think it should be fixed now with the commit. legal.csv and legal-eu.csv have data as expected, and legal.csv uses the correct vocab and namespace. Copied over sample outputs from legal.csv:

Also checked for regression: dpv.csv, pd.csv, and eu-gdpr.csv have correct data.

from dpv.

@thovden please see CSV files - where:

• term is the DPV term without vocab namespace e.g. AcademicResearch
• type represents if it is a rdfs:Class or rdf:Property
• dpvtype is the DPV term whose instance the term is e.g. Purpose
• subclassof indicates the DPV term that is used with rdfs:subClasOf
• hasbroader lists the DPV terms used with skos:broader
• vocab provides the vocab identifier e.g. dpv
• namespace provides the vocab IRI e.g. https://w3id.org/dpv
• iri = namespace + term

from dpv.

[@coolharsh55] @thovden please see CSV files - where:

I've edited the following, to better present the info —

• term is the DPV term without vocab namespace, e.g., AcademicResearch
• type represents whether it is an rdfs:Class or rdf:Property
• dpvtype is the DPV term whose instance the term is, e.g., Purpose
• subclassof indicates the DPV term that is used with rdfs:subClasOf
• hasbroader lists the DPV terms used with skos:broader
• vocab provides the vocab identifier, e.g., dpv
• namespace provides the vocab IRI, e.g., https://w3id.org/dpv/#
• iri = namespace + term, e.g., https://w3id.org/dpv/#AcademicResearch

(First off, either namespace above must be changed as I have, i.e., https://w3id.org/dpv must become https://w3id.org/dpv/#, or iri above must change to `namespace` + `/#` + `term`. And, if all this data is being managed in GSheets, all those descriptions should be incorporated in that worksheet, either as cell comments or (probably better) as a new row.)

I can't imagine that there are many humans who find CSV files useful for manual review. Usually, CSV files are used for data transfer, and loaded/exported to/from applications that offer tabular editing interfaces, similar to what you have presented here.

GitHub is a terrible place in which to manage data. It's simply not designed for it. Changes, suggestions, etc., do need to be tracked, and credit/blame extended to the relevant authors, editors, etc. Google Spreadsheets might be the best option, today, though it will mean that extra effort needs to be applied by the spreadsheet owner, to map Google User IDs to GitHub Handles, among other things (such as documenting import/export procedures, such that all editors execute the same steps and don't break each other's efforts — which is difficult to prevent, unless all editors are working on the same worksheet instance, and doing that work within GSheets)....

from dpv.

@TallTed this is the output CSVs being produced as another format/serialisation alongside RDF and HTML. The first comment in the issue clearly states this. Google sheets spreadsheets are where we do discussions on new concepts and changes - that is not linked to this work at all so nothing needs to be mentioned/put there from this issue.

from dpv.

For IRIs and namespaces - will the exported RDF, CSV and so on have the specific versioned IRIs - e.g., https://w3id.org/dpv/v2.0/ai#AI or the unversioned one https://w3id.org/dpv/ai#AI or both? If the definitions and/or semantics change across versions we need to track which version of a term is in use in our application.

Also see #45

from dpv.

@coolharsh55 looking good for the most part on my side. However, the legal-eu vocab is empty in the CSV file, and e.g., the adequacy decision are not there. They are exported into the other formats.

from dpv.

@thovden this should be fixed now - please see https://github.com/w3c/dpv/blob/dev/v2.0/legal/legal.csv

from dpv.

@coolharsh55 legal-eu.csv is still empty https://github.com/w3c/dpv/blob/dev/v2.0/legal/eu/legal-eu.csv. As long as these are pulled into the legal vocab I guess I'm good.

from dpv.

In https://w3c.github.io/dpv/legal/ it states that there is a legal-eu vocab for e.g., the adequacy decision. However, with the latest CSV export the vocab is eu-gdpr. The namespace is https://w3id.org/dpv/v2.0/legal/eu/gdpr (which is also reflected in the directory structure).

There are also examples where the IRI is e.g., https://w3id.org/dpv/v2.0/legal/eu#Adequacy-EU-FO and the namespace is https://w3id.org/dpv/v2.0/legal. I would expect vocab to to legal-eu in this case and not legal.

I think it would be nice to have a consistent mapping (and all the namespace bits) between IRI and the vocab:term for the short form. So for example: IRI: https://w3id.org/dpv/v2.0/legal/eu/gdpr#A6-1-b-contract-performance could be:

• vocab: legal-eu-gdpr (alternatively legal:eu:gdpr)
• term: A6-1-b-contract-performance
• short form: legal-eu-gdpr:A6-1-b-contract-performance (alternatively legal:eu:gdpr:A6-1-b-contract-performance)

Without this consistent mapping we cannot automatically extract the vocab from the IRI. I guess I'm a bit late to the party here, so I understand if this is difficult now, but just wanted to flag this.

from dpv.

@thovden we had to change the IRI (again) as our implementation was not according to best practice. We removed versioned IRIs e.g. https://w3id.org/dpv/2.0#Purpose and replaced it with the versionless IRIs e.g. https://w3id.org/dpv#Purpose. Please review the CSV.

The term IRIs will not change across different versions. The entire ontology/file will be available at the versioned IRIs e.g. https://w3id.org/dpv/2.0 will give DPV v2 with terms using the versionless IRI.

from dpv.

Confirmed via DM/emails the CSV outputs are okay. Closing issue as completed.

from dpv.