This is the archive of the goodrelations dicussion list

Hi Ed,

> Is it better to have the sameAs reference a parallel canonical URI -
> like a SPARQL endpoint - unrelated to the website topography (is this
> option #2?).

	In general, I recommend to make hash URIs based on URIs of the human- 
readable HTML page the authoritative ones, as long as the pages  
contain RDFa markup.
By that, you always refer someone who is trying to dereference a data  
entity (e.g. an offer or a product) to a resource that can serve both  
human-readable and machine-readable content.

Even if you provide data dumps (e.g. in RDF/XML) or SPARQL endpoints,  
I usually suggest to use the URI references derived from the  
authoritative Web pages.

It is essential to understand that this design choice influences the  
URI which will be tried for fetching information about the data object.

So if you serve both a data dump in RDF/XML and RDFa in shop pages, I  
suggest to use the URIs resulting from parsing the RDFa also for the  
RDF/XML data.

The HTML+RDFa and RDF/XML templates at

   http://code.google.com/p/templates4goodrelations/

also follow that approach. If interested in the details, please  
compare the output of the data dump file with the individual page  
markup.

> Or is there a way to publish all product URI with sameAs
> patterns such that the first discovered is taken as the canonical URI
> for any following duplicate URI? The objective would be to remove the
> web of maintenance from using the sameAs pattern.

As I tried to express, the sameAs approach is the weakest, for it will  
still require crawling the same content from multiple URIs in order to  
consolidate all URI variants.
Still, it is much better than nothing, because otherwise, a dataspace  
operator would have to rely on imperfect heuristics for entity  
consolidation.

> Thanks for all the GoodRelations  - Ed
Thanks!

Best

Martin

On 24.08.2010, at 20:12, Ed - 0x1b, Inc. wrote:

> On Tue, Aug 24, 2010 at 6:04 AM, Martin Hepp
> <martin.hepp at ebusiness-unibw.org> wrote:
>> Dear all:
>>
>> Some shop applications, unfortunately, display the very same item at
>> multiple URIs. This is problematic for Search Engine Optimization
>> (SEO) and the Web of Data alike.
>>
>
> regarding:
>
>> 3. The last option (but the least powerful, yet still much better  
>> than
>> doing nothing) is to add owl:sameAs statements from the RDFa pattern
>> to the canonical URI.
>> The canonical URI should also be used for the foaf:page property,
>> which is the crucial link from the data to the page from where the
>> product can be ordered.
>>
>> Example:
>>
>>   <div about="#offering" typeof="gr:Offering">
>>     <div rel="owl:sameAs" resource="http://www.myshop.com/staplers/green_pocket_stapler123#offering
>> "></div>
>>     <div property="rdfs:label" content="Cool green stapler for $8.99"
>> xml:lang="en"></div>
>> ...
>>     <div rel="foaf:page" resource="http://www.myshop.com/staplers/green_pocket_stapler123
>> "></div>
>>   </div>
>> </div>
>>
>
> Is it better to have the sameAs reference a parallel canonical URI -
> like a SPARQL endpoint - unrelated to the website topography (is this
> option #2?). Or is there a way to publish all product URI with sameAs
> patterns such that the first discovered is taken as the canonical URI
> for any following duplicate URI? The objective would be to remove the
> web of maintenance from using the sameAs pattern.
>
> Thanks for all the GoodRelations  - Ed