W3C home > Mailing lists > Public > public-vocabs@w3.org > January 2015

RE: FYI: Interval Vocabulary

From: <Simon.Cox@csiro.au>
Date: Wed, 28 Jan 2015 05:31:47 +0000
To: <jasnell@gmail.com>
CC: <public-vocabs@w3.org>
Message-ID: <2A7346E8D9F62D4CA8D78387173A054A600BC651@exmbx04-cdc.nexus.csiro.au>
> Ultimately, however, it would likely just be easier to use domain specific definitions for things like bounding boxes rather than rely on this more generalized approach.

This will no doubt be on the agenda in the SDW-WG. But note that the most rigorous definition of 'bounding box' (called 'envelope') in ISO 19107 uses exactly this approach - a lower and upper corner. Same count of numbers, as the more common bounding-box models, but more general as it is not tied to a specific topological dimension. 

Simon 

-----Original Message-----
From: James M Snell [mailto:jasnell@gmail.com] 
Sent: Saturday, 17 January 2015 10:55 AM
To: Cox, Simon (L&W, Highett)
Cc: public-vocabs@w3.org Vocabs
Subject: Re: FYI: Interval Vocabulary

Yes, but unfortunately it depends largely on how the geo point is defined. In order for the Interval to be meaningful, the value space must be ordered. For now I've limited upper/lower to xsd datatypes that with the ordered facet given. A geo point is a bit more difficult but not impossible... an implementation would simply need to be aware of how exactly to determine the ordering of the points. I have considered, however, how we could incorporate arbitrarily complex objects by allowing the Interval to identify specific order keys. e.g.

_:c14n0 a :ClosedInterval ;
  :orderedBy (
    [ a :AscKey ; :ref <geo:long> ],
    [ a :DescKey ; :ref <geo:lat> ]
  ) ;
  :lower [
    a geo:Point ;
      geo:lat "85.200000"^^xsd:float ;
      geo:long "-90.300000"^^xsd:float .
  ] ;
  :upper [
    a geo:Point ;
      geo:lat "80.200000"^^xsd:float ;
      geo:long "-85.300000"^^xsd:float .
  ] .

This is just a strawman, of course, and it adds a fair amount of complexity, but it demonstrates how things can work. With this, one would still only need to understand ordering of the simple data types.
Ultimately, however, it would likely just be easier to use domain specific definitions for things like bounding boxes rather than rely on this more generalized approach.

On Wed, Jan 14, 2015 at 10:18 PM,  <Simon.Cox@csiro.au> wrote:
> And if you use a geo Point for the type of upper/lower, then you get bounding boxes (in any dimension).
>
> -----Original Message-----
> From: James M Snell [mailto:jasnell@gmail.com]
> Sent: Thursday, 15 January 2015 2:38 PM
> To: public-vocabs@w3.org Vocabs
> Subject: Fwd: FYI: Interval Vocabulary
>
> FYI...
>
> ---------- Forwarded message ----------
> From: James M Snell <jasnell@gmail.com>
> Date: Wed, Jan 14, 2015 at 4:55 PM
> Subject: FYI: Interval Vocabulary
> To: "public-socialweb@w3.org" <public-socialweb@w3.org>
>
>
> Just wanted to make folks aware of this..
>
>   http://ns.jasnell.me/interval

>
> It's a small vocabulary for representing bounded intervals (totally ordered sets). It is born out of various Activity Streams use case including the ability to indicate validity date ranges for certain types of objects, identifying filter ranges for Collection items and other wonderful things.
>
> It ought to be fairly straightforward. Would eventually be interested in seeing this become a WG Note but definitely not a priority.
>
> - James
>
Received on Wednesday, 28 January 2015 05:32:29 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 28 January 2015 05:32:30 UTC