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

RE: Meta -> data use case/API

From: Tandy, Jeremy <jeremy.tandy@metoffice.gov.uk>
Date: Thu, 3 Apr 2014 13:15:52 +0000
To: Phil Archer <phila@w3.org>
CC: "public-csv-wg@w3.org" <public-csv-wg@w3.org>
Message-ID: <2624871D9A05174691BD59F8EFD68AE208833892@EXXCMPD1DAG3.cmpd1.metoffice.gov.uk>
I certainly agree that a CSV profile of OData would be useful ... with the Sensors and Observations use case I mooted, I can see the compact CSV encoding being idea for situations where one has many (tens of) thousands of observation instances to publish!

Agreed that the hook we need is metadata -> data@URI ... AndyS's point about annotating a subset of the data@URI is also valid. 

Now all we need is a "real" use case to hang this on. Are you still content to provide this Phil?


-----Original Message-----
From: Phil Archer [mailto:phila@w3.org] 
Sent: 03 April 2014 14:00
To: Tandy, Jeremy
Cc: public-csv-wg@w3.org
Subject: Re: Meta -> data use case/API

When getting the WG set up, I had an extended exchange with Microsoft over the possibility of (them) introducing a CSV profile of OData. I think it's something they may want to do in due course. It's obviously up to them and the other parties interested in OData but the hook for us is that if we have that metadata -> data URI|L then it remains an option.

I didn't mean this WG to be deflected into considering OData directly as long as it remains JSON/ATOM only.



On 03/04/2014 13:46, Tandy, Jeremy wrote:
> Phil - regarding OData, further inspection indicates that the spec requires JSON or ATOM encoding of the data ... it's not clear to me how we'd role that into a use case talking about tabular text (CSV) data.
> Jeremy
> -----Original Message-----
> From: Tandy, Jeremy [mailto:jeremy.tandy@metoffice.gov.uk]
> Sent: 02 April 2014 21:24
> To: Gregg Kellogg; Phil Archer
> Cc: public-csv-wg@w3.org
> Subject: RE: Meta -> data use case/API
> Gregg, Phil.
> Thanks for starting this thread. I think this is an important case to 
> cover (& currently absent) ... I have two angles in mind that might be 
> worth pursuing for a usecase
> #1: there is a considerable amount of CSV already published, such as on open data portals like data.gov.uk; the majority of this is not annotated and is unlikely to be until the publishers "get around" to retrofit the data with metadata annotations. In the real world, this could be never. Therefore, it is important that third parties can add the annotation to complement data for which they are not the publisher. UC #17 "Canonical mapping of CSV" <http://w3c.github.io/csvw/use-cases-and-requirements/#UC-CanonicalMappingOfCSV> already states "there is nothing stopping a data publisher from adding annotation defining data semantics once, say, an appropriate de facto standard vocabulary has been agreed within the community of use", so this might provide a hook for incorporating the "metadata refers to data published by a third party" requirement.
> #2: I came across standards initiative in the Open Geospatial Consortium (SensorThings API), as yet incomplete, that uses the OData protocol to publish observations as datastreams from internet connected sensors (in the theme of Internet of Things). I'll dig a bit further to see if this offers a use case for us to annotate data made available via the OData protocol (at the moment, it looks quite json centric and does not appear to offer a CSV encoding).
> Gregg's idea about annotating a html table published by a 3rd party 
> might be simpler to communicate :)
> -----Original Message-----
> From: Gregg Kellogg [mailto:gregg@greggkellogg.net]
> Sent: 02 April 2014 16:37
> To: Phil Archer
> Cc: Tandy, Jeremy; public-csv-wg@w3.org
> Subject: Re: Meta -> data use case/API
> On Apr 2, 2014, at 7:43 AM, Phil Archer <phila@w3.org> wrote:
>> Jeremy,
>> Picking up on your +1 to my mention of a use case where person B creates metadata about data published by person A, I'm happy to write the use case(s) but do you have one in mind that you can point me to? As well as this general case, it should also cover metadata for a query to an API (potentially including OData), i.e. anything that comes back from a URL that looks a bit like a table. It's the link *from* the metadata *to* the data that I'm most anxious to see covered.
> It seems to me that allowing the metadata to reference the data source should be fairly straightforward. The case where this is a URL would seem to cover this, and we would want to provide guidance on content-negotiation for this. We also need to provide some enumeration of specific supported types (text/csv, text/tsv). It might be interesting to consider extracting data from an HTML table identified by a fragment of an HTML document as another supported format.
> Gregg
>> Phil.
>> --
>> Phil Archer
>> W3C Data Activity Lead
>> http://www.w3.org/2013/data/
>> http://philarcher.org
>> +44 (0)7887 767755
>> @philarcher1


Phil Archer
W3C Data Activity Lead

+44 (0)7887 767755
Received on Thursday, 3 April 2014 13:16:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:21:39 UTC