[Developers] rdf auto-discovery and dbpedia

[Kingsley Idehen]

Ed Summers wrote:
> On Mon, Dec 1, 2008 at 5:23 PM, Scott Meyer <sm at metaweb.com> wrote:
>   
>> Suppose, for example, that there is another space topic which "means"
>> the same thing.  While we try and prevent such duplication from occurring,
>> it can happen.  When we discover duplication we pick one topic as the winner
>> (usually the one with the most data) and merge all data *and keys* from the
>> loser to the winner.
>>     
>
> In this case wouldn't you also have a "key" URL that incorporated the
> name of the other 'space topic' such as: en/outer-space?
>
>   
>> So, if your code refers to "/en/space" it will "just work" after a merge
>> but if you refer to guids directly your code will have to check for
>> topics which have been merged and follow a link to the new topic,
>> if necessary.
>>     
>
> Assuming that I am right about en/outer-space, what would happen to it
> when it was merged into en/space? Would /en/outer-space 404 like the
> GUID based identifier?
>
> In general I don't see how the problem of two identifiers for the same
> thing is solved with 4 identifiers for the same thing. But perhaps I'm
> missing something.
>
> //Ed
> _______________________________________________
> Developers mailing list
> Developers at freebase.com
> http://lists.freebase.com/mailman/listinfo/developers
>
>   
Ed,

Freebase will have to at least make the "owl:sameAs" assertions as data 
space owners. Once this is done, inferencing will take care of the rest. 
Of course, inferencing cost and quality will be inference engine 
dependent etc.. 


-- 


Regards,

Kingsley Idehen	      Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO 
OpenLink Software     Web: http://www.openlinksw.com