Re: rdfa vs. links

Hi William,

As long as you are re-using existing subject URIs with the "about"  
attribute or provide strong links to the original entity URIs, it is  
of course fine to replicate only part of the data in RDFa.

But it is crucial to NOT create unconnected, redundant entity  
definitions in RDFa.

One argument in favor of richer RDFa is that client-side applications  
will often just use what is directly available in the page and not  
request external resources, even if properly referenced.

Personally I think that RDFa as a generic vehicle for RDF is a very  
powerful approach, because

1. it requires minimal skills re server configuration
2. it can be created using HTML templating engines and frameworks that  
are widely understood by Web developers.

I even created Djnago-style templates for generating GoodRelations RDF/ 
XML [1, also contains powerful RDFa templates] so that a Web developer  
does not need any RDF or XML library to produce rich RDF data dumps. I  
know that is ugly for purists, but HTML templating engines are likely  
the most widely comprehended technique for creating Web content.

Martin

[1] http://code.google.com/p/templates4goodrelations/
On 26.10.2010, at 18:11, William Waites wrote:

> On Tue, Oct 26, 2010 at 03:31:32PM +0100, Michael Hausenblas wrote:
>>
>> Your question motivated me to eventually carry out an experiment I  
>> thought
>> of a while ago. Now I got my act together and eventually published  
>> some
>> observations on the topic of HTML+RDFa load time dependencies on  
>> the number
>> of embedded triples [1]. I hope you find this useful ;)
>
> Hi Michael,
>
> I guess my question was more *should* than *how* to use
> links to the RDF vs. embedded RDFa. Your post points to
> an upper bound for where should turns into shouldn't (at
> some number of triples where the page takes too long to
> load).
>
> As far as should one put RDFa to begin with, supporting
> cut-and-paste operations in the UI seems like a compelling
> reason (caveats about implementations properly getting
> the subject URI aside).
>
> Somehow I'm still leaning towards a minimalistic position
> only putting as much information in RDFa as necessary to
> give enough context to pull the entire resource. For
> example if I mention Richard Feynman I might want to
> embed dbpedia:Richard_Feynman a foaf:Person but not include
> copious biographical information in hidden <span>s (unless
> I'm writing his biography).
>
> Cheers,
> -w
>

Received on Tuesday, 26 October 2010 16:23:52 UTC