W3C home > Mailing lists > Public > public-csv-wg@w3.org > September 2014

Re: (Updated) draft for a default RDF mapping plus a draft for JSON mapping

From: Ivan Herman <ivan@w3.org>
Date: Mon, 29 Sep 2014 06:33:09 +0200
Cc: W3C CSV on the Web Working Group <public-csv-wg@w3.org>
Message-Id: <8199741B-EFD0-4013-82EA-C321ECF03D3F@w3.org>
To: Yakov Shafranovich <yakov-ietf@shaftek.org>
Hey Yakov,

On 29 Sep 2014, at 03:36 , Yakov Shafranovich <yakov-ietf@shaftek.org> wrote:

> Ivan - thanks for the great work on this.
> 
> I am wondering as another experiment if it would be feasible to come
> up with an XSLT template example that can then use the RDF result for
> something else. This would give some teeth to at least one of the
> template alternatives we are discussing.
> 

My instinctive reaction is: yes an no:-)

Yes, in the sense that it would indeed show the possibility to use the output of the simple output combined with a further transformation engine to show that use cases can be covered that way. So this should indeed be part of the experiment. Which use case would be the most convincing for this? Not sure, maybe Jeremy could tell.

No, in the sense that I would keep away from XSLT in this context. First of all, the default mapping is in terms of abstract RDF triples; using XSLT would presuppose to serialize the triples in RDF/XML. And, personally, I would really like keep away from RDF/XML as much as I can... I think that using a SPARQL CONSTRUCT pattern (or maybe SPIN) would be more appropriate.

B.t.w.: is there a widely accepted and used transformation engine for JSON? What would one use for that? I can of course imagine a Javascript program doing pretty much everything, so this may be o.k. for Javascript based implementation...

Thanks

Ivan 



> Thanks,
> Yakov
> 
> On Thu, Sep 25, 2014 at 9:00 AM, Ivan Herman <ivan@w3.org> wrote:
>> As agreed on our call, I have updated my earlier draft for a default RDF mapping. This is now on:
>> 
>> http://w3c.github.io/csvw/csv2rdf/
>> 
>> As an experiment, I have also created a copy of it where I changed RDF to JSON, trying to define a minimal JSON mapping. The result is:
>> 
>> http://w3c.github.io/csvw/csv2rdf/index-json.html
>> 
>> The goal is to keep these very close; ideally, if one adds a suitable @context then the JSON one should be a serialization (in JSON-LD) of the RDF mapping. I am sure it is not proper, and I hope that somebody will look at that when the time comes.
>> 
>> I have added a number of issues to the document (and there are probably more). The biggest issue I have hit is as follows. At present, we may find ourselves in a situation when no column name is specified. If the CSV file is a 'Core Tabular Data' (but not annotated), and the metadata does not have a schema/column data set; all these are indeed optional. In a previous version of the data model there was a rule set on some sort of a default naming of the columns, but that is, somehow, gone. I have added something in this document, but I am not sure that is the right place.
>> 
>> Enjoy:-)
>> 
>> Cheers
>> 
>> Ivan
>> 
>> P.S. Tracker, this closes Action-30
>> 
>> ----
>> Ivan Herman, W3C
>> Digital Publishing Activity Lead
>> Home: http://www.w3.org/People/Ivan/
>> mobile: +31-641044153
>> GPG: 0x343F1A3D
>> WebID: http://www.ivan-herman.net/foaf#me
>> 
>> 
>> 
>> 
>> 


----
Ivan Herman, W3C 
Digital Publishing Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
GPG: 0x343F1A3D
WebID: http://www.ivan-herman.net/foaf#me






Received on Monday, 29 September 2014 04:33:39 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:21:41 UTC