Shared publicly  - 
 
A call to Google to make the next step towards integration of the Semantic Web
4
5
Luca Matteis's profile photoBernard Vatant's profile photo
10 comments
 
"Getting URIs of things as part of a regular search from the major search engine would be a significant milestone. " but isn't that already how Google works? Google works with URLs (therefore URIs)... the more a URL is found inside pages, the more ranking it'll get in Google's result. I don't understand what changes exactly you're expecting...
 
But if those dbpedia links were relevant (were found many times across pages indexed by search-engines), than they would be shown as a result, just like regular pages. I don't see why you think that an RDF link is any different than any other link. It's not! That's why RDF uses URI's for identifiers.... so that technology already built around pages (search-engine's scraping algorithms) can be used for these pieces of information as well.
 
+Luca Matteis OK let me put it otherwise for you : Google displays results for images, or videos, or news differently from general pages, right? Results for URIs-of-the-thing should be distinguished the same way. Among results, you would have pages URI, and images URI, and video URIs, and news URI, and thing-itself URIs. Does that make sense?
 
Sorry I still don't get it :( If you build your RDF page correctly (using RDFa for example) you CAN benefit from google's rich-snippets and other goodies. Therefore RDF links could very well be displayed differently, just as Images or Videos are... it all depends on the RDF provider - they need to do a good in job in marking up their html into RDFa. For the DBPedia example, in my opinion that's Dbpedia's fault. There should only be one link and not two links for the same thing. Maybe I can ask you a question: what is exactly that you want search-engines to do? Say you Google for "Victor Hugo" on Google - what would it need to display in order to satisfy your request?
 
I can't explain it more or better than above, sorry. If you don't get my point, let's agree to disagree then :)
 
This is the internet. There's no such thing ;)
 
+Luca Matteis Not sure what your last cryptic sentence means.
Two answers (plus one) since you insist :
1. About DBpedia URIs being "wrong" to have "two links for the same thing", please read all that has been written over years about URIs of things vs URIs of pages, resolution of httpRange14 etc. to understand the thing/page distinction.
If DBpedia is wrong on this, so are the W3C and Tim Berners-Lee himself, followed by the Library of Congress and all major libraries supporting VIAF, data.gov, data.gov.uk, data.gouv.fr, Europeana, FAO, New York Times and BBC, Geonames, INSEE and IGN in France .. and many more. Can Google keep on ignoring the work of all those actors? Aren't they part of the Web?
2. About what Google could display : sorting the 70 URIs at http://sameas.org/html?q=Victor+Hugo, which ones are really representing Victor Hugo, which ones are the most used and trustable etc. and display them as specific results "URI identifying Victor Hugo on the Web".
3. And Google is just an example ... Bing and Yahoo! are not better in this respect. I focused on Google because with the Knowledge Graph one could think that it has groked the basics of linked data :)
 
All I'm saying is that it would make more sense if we had a single URI for the "thing" and for the "page". I mean, what are the benefits of having separate URIs? The whole point of using URIs as IDs is that they are referenceable. But overall I see your point and I agree that Google could do a better job to help with Semantics.
Add a comment...