Comments on: The Universal Data Model http://netzooid.com/blog/2008/02/10/the-universal-data-model/ gettin all zoidal on ya Sat, 24 Jul 2010 09:48:52 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Stu http://netzooid.com/blog/2008/02/10/the-universal-data-model/comment-page-1/#comment-30421 Stu Wed, 13 Feb 2008 21:31:40 +0000 http://netzooid.com/blog/2008/02/10/the-universal-data-model/#comment-30421 Egads, I didn't escape my XML there. Here was the example: <set name="stuff"> <row> <column key="true" name="a">value1</1> <column name="a">value1</1> <column name="a">value1</1> </row> </set> or <stuff key="a"> <a>value1</a> <b>value2</b> <c>value3</c> </stuff> Egads, I didn’t escape my XML there. Here was the example:

<set name="stuff">
<row>
<column key="true" name="a">value1</1>
<column name="a">value1</1>
<column name="a">value1</1>
</row>
</set>

or

<stuff key="a">
<a>value1</a>
<b>value2</b>
<c>value3</c>
</stuff>

]]> By: Stu http://netzooid.com/blog/2008/02/10/the-universal-data-model/comment-page-1/#comment-30420 Stu Wed, 13 Feb 2008 21:28:40 +0000 http://netzooid.com/blog/2008/02/10/the-universal-data-model/#comment-30420 Regarding "pushing the problem elsewhere": The whole point is that you want to be able to break problems down with appropriate levels of indirection. It's the whole point behind abstraction, and it's a major driver for RESTful architectures. For example, there is a classic tradeoff between generic representations and specific representations. In XML, I could say value1 value3 value3 or <b>value2</b> value3 There are pros and cons to either approach. From one POV, RDF is a well specified case of the former, and raises interoperability to the level of logic instead of just syntax, like what we (try, sometimes) do with relational databases. OTOH, I wouldn't want to transfer or interpret bitmap images as RDF triples. So, if I want to associate metadata with the JPG, either I: 1. provide a wrapped representation with another format, say, an Atom entry (Nick Gall's suggestion in my blog comments), or more commonly, HTML and Microformats 2. provide an RDF content negotiated representation for the image, see <a href="http://www.w3.org/TR/photo-rdf/" rel="nofollow">this note</a> as an example 3. add an HTTP header or parameter to Content-Type (this debate) 4. modify JPG itself to enable URI hooks (not bloody likely) These probably could be generalized to any media type that allows for extensions or mixed vocabularies. Regarding “pushing the problem elsewhere”: The whole point is that you want to be able to break problems down with appropriate levels of indirection. It’s the whole point behind abstraction, and it’s a major driver for RESTful architectures.

For example, there is a classic tradeoff between generic representations and specific representations. In XML, I could say

value1
value3
value3

or

value2
value3

There are pros and cons to either approach. From one POV, RDF is a well specified case of the former, and raises interoperability to the level of logic instead of just syntax, like what we (try, sometimes) do with relational databases.

OTOH, I wouldn’t want to transfer or interpret bitmap images as RDF triples.

So, if I want to associate metadata with the JPG, either I:
1. provide a wrapped representation with another format, say, an Atom entry (Nick Gall’s suggestion in my blog comments), or more commonly, HTML and Microformats
2. provide an RDF content negotiated representation for the image, see this note as an example
3. add an HTTP header or parameter to Content-Type (this debate)
4. modify JPG itself to enable URI hooks (not bloody likely)

These probably could be generalized to any media type that allows for extensions or mixed vocabularies.

]]>