[Geojson] Interesting comparison of JSON and RDF

[chris goad]

Hi Allan, Sean,

>> I've never felt that GeoJSON should be put forward as a general
>> purpose
>> interchange format, and that's why I've resisted the RDF-isms of JDIL.
>


I wasn't proposing that GeoJSON itself be formulated as a general purpose 
format, but rather that minor GeoJSON details should adjusted to allow its 
use in a more general context. This is analogous to the GeoRSS GEO 
primitives being defined in such a way that they can appear within an RSS 
feed - allowing people who want to make geographical assertions about their 
content to do so even if their content mentions things other than geography. 
Concretely speaking, at Platial we use JSON as a basis for sending various 
kinds of information around (commentary, photos, specialized data about 
particular topics), and need something that we can mix in to carry the geo 
content.


In reply to Allan's comments:


The art of mixing data from different sources is not arcane, nor is it 
intrinsically tied to notions of  semantic nirvana - though it may seem to 
be to listen to some hype. The technical notions are trivial: 1) avoid name 
collisions by using long names (URIs) instead of short ones; 2) to deal with 
the resulting versbosity, define abbreviations (namespaces). This is old hat 
in the XML world, and JDIL is an attempt to transplant this simple but 
useful hack into JSON.


As evidence that data-mixing and  namespaces are not just an affectation of 
the semantic faithful, the Google Data APIs which provide access to Google 
Base, Calendar, Spreadsheets, Picasa  Web Albums etc use namespaces-per-app 
to mix the data from these applications into Atom or JSON feeds.  (If that's 
not mainstream what is?)


GEO in particular can benefit hugely from data mixing - geographic features 
always have non-geographic attributes that are of central interest to one 
community or another : the strength as well as position of earthquakes, the 
menu as well as address of a restaurant.  IMHO, Geo format designers  should 
consider how  best to fit into more general information flows rather than 
how best to sequester their format in its purity!


BTW, we will be using GeoJSON at platial - but might need to sully it  with 
the occaisonal  "@"


Final note: I commented on 
http://blogs.sun.com/bblfish/entry/the_limitations_of_json - trying to show 
that JSON  isn't limited after all for the purpose of transporting RDF - if 
one should wish to do that. Thanks for posting  the link.


-- Chris





----- Original Message ----- 
From: "Allan Doyle" <adoyle at eogeo.org>
To: <geojson at lists.geojson.org>
Sent: Thursday, July 26, 2007 2:30 PM
Subject: Re: [Geojson] Interesting comparison of JSON and RDF


>
> On Jul 26, 2007, at 11:13 , Sean Gillies wrote:
>
>> You may have read this already, else open it in a new tab and read it
>> when you get a chance.
>>
>> http://blogs.sun.com/bblfish/entry/the_limitations_of_json
>>
>> I've never felt that GeoJSON should be put forward as a general
>> purpose
>> interchange format, and that's why I've resisted the RDF-isms of JDIL.
>
> I agree that GeoJSON isn't targeted at being a general purpose
> interchange format. However, I also cringe every time I read
> something like that blog posting.
>
> People keep thinking that if they only add a little more notation,
> they will achieve semantic nirvana where there is no out-of-band
> information needed and everything can just grok everything else.
> Sure, adding namespaces like foaf: helps you along the way, but how
> do people ever get anything done once they start indirecting
> everything? It seems to me that the trend is towards late-binding so
> that one doesn't have to commit immediately to nailing things down.
> To truly achieve a semantic web, the binding may have to be
> infinitely late. In the meantime, the early-binding constructs are
> out there doing useful things.
>
> I think GeoJSON as is, or maybe even with the feature e stuff pulled
> out is a fine piece of work and we should not be lulled into thinking
> that just one more little bit of notation won't be so bad.
>
> Maybe it's because I started out life as a hardware designer. I feel
> that when you're done, you should have something pretty self-
> contained that just works, preferably in a deterministic way. Or
> maybe I've ossified to the point where I just don't get it and the
> young-uns will have to just move ahead without me.
>
> Allan
>
>>
>> Sean
>> _______________________________________________
>> Geojson mailing list
>> Geojson at lists.geojson.org
>> http://lists.geojson.org/listinfo.cgi/geojson-geojson.org
>
> -- 
> Allan Doyle
> +1.781.433.2695
> adoyle at eogeo.org
>
>
>
> _______________________________________________
> Geojson mailing list
> Geojson at lists.geojson.org
> http://lists.geojson.org/listinfo.cgi/geojson-geojson.org