Re: Newbie frustrations

On 4 Jan 2006, at 18:53, wollman+semantic-web@bimajority.org wrote:

> <<On Wed, 4 Jan 2006 18:25:18 +0100, Richard Cyganiak  
> <richard@cyganiak.de> said:
>
>> Here's another thought about your points c) and d) and the  
>> dc:creator/
>> foaf:Person issue: You say you want to add metadata to "assist a
>> search application". That's pretty vague. What kind of functionality
>> do you want to enable?
>
> Well, in the general case, I want to enable whatever users of my
> gallery builder (if there are any other users) want to do with their
> metadata; that motivates the full generality of what I'm trying to do.

Sure, that's what anyone should strive for. At least personally, I  
find that if I try to solve a modelling problem with any possible use  
of the data in mind, then I tend to collapse from lack of oxygen  
after a while. Hence the "make up some scenario for the data  
consumption side" approach.

> In the more specific case, I, as a user of my gallery builder, want to
> answer queries of the form "show me all the information related to X"
> where X is an externally-assigned identifier.  Anything else is well
> beyond the scope of what I'm currently trying to do -- but given the
> amount of manual work involved in annotating thousands of photos, I
> want to get the structure right from the start so I don't have to
> repeat all that work if I decide to add additional functionality
> later.

Yes, that makes a lot of sense. The safe route might be not to commit  
to any particular RDF representation, but just to record bits of  
metadata that are guaranteed to be useful *for your case*. You're  
working off an XML file, so maybe you could add something like this  
to every photo where it's applicable:

    <depicted-antenna-number>123</depicted-antenna-number>

The concrete RDF representation could be generated as part of the  
XSLT transform. I like the representation given by Richard (Newman)  
in a previous answer, but you might pick another one later if  
experience suggests a better one.

This is a *lot* less flexible than a generic "Here goes RDF metadata"  
field, but it is dead simple, solves your concrete problem, and can  
be extended easily to cover other cases. It doesn't solve the lack-of- 
oxygen inducing general case though.

Thoughts?

Richard


> The path of least resistance would be simply to annotate each
> photo with the property "subject has number X", but this is deeply
> unsatisfying: one would rather say "subject is antenna which has
> number X".
>
> -GAWollman
>
>

Received on Wednesday, 4 January 2006 19:28:45 UTC