Re: Time for quintuples?

Hi William,

thanks a lot for your insightful answer.

more comments inline ;)

Am 28.09.2010 11:57, schrieb William Waites:
> On 10-09-26 23:09, Bob Ferris wrote:
>>
>> ex:ACC a cco:CognitiveCharacteristic ;
>>      cco:agent ex:ww ;
>>      cco:characteristic cco:belief ;
>>      cco:topic ex:ATopic ;
>>      wo:weight [
>>          a wo:Weight ;
>>          wo:weight_value 0.12 ;
>>          wo:scale ex:AScale
>>          ] ;
>>      cco:activity<http://dbpedia.org/resource/Category:Thinking>  .
>>
>> ex:ATopic {
>>      ex:bob cco:skill<http://dbpedia.org/resource/Football_(soccer)>
>> }    # here the semantics are getting a bit arguable
>>
>> ex:AnotherCC a cco:CognitiveCharacteristic ;
>>      cco:agent ex:bob ;
>>      cco:characteristic cco:skill ;
>>      cco:topic http://dbpedia.org/resource/Football_(soccer)>  ;
>>      cco:weight [
>>          a wo:Weight ;
>>          wo:weight_value 0.06 ;
>>          wo:scale ex:AScale
>>          ] ;
>>      cco:activity
>> <http://sw.opencyc.org/concept/Mx4rwJRiEpwpEbGdrcN5Y29ycA>  .
>>
>> ex:AScale a wo:Scale ;
>>     wo:min_weight 0.0 ;
>>     wo:max_weight 9.0 ;
>>     wo:step_size 0.1 .
>>
>> Would you agree with that modelling?
>
> Yes, it is a reasonable approach. I don't disagree with it.
>
> I think that notation-wise, perhaps nested graphs are a
> bit clearer to read than using reification. It might be worth
> considering why you had to make resource to referring to a
> graph for the object of a belief. And if you start talking
> about beliefs about beliefs maybe the shape will start being
> more and more like nested graphs.
>

Yes, it's simply for reusing all these descriptions in a distributed 
interlinked knowledge base.

> OTOH nested graphs might be controversial and are not well
> supported.

Yes, that's the first reason, why I prefer URIing everything  explicitly 
to make these descriptions reusable.

>
> (I also don't particularly see why graphs, (ex:ATopic) have to
> be explicitly named, they could just as well be anonymous (named
> with a blank node) but that's a different discussion.)

In a scenario with a distributed interlinked knowledge base, I came to 
the conclusion to provide URIs for everything and avoid to use blank nodes.
I came to the conclusion that blank nodes are probably good in a closed 
dataset and for demonstrating use cases, but they are bad when doing 
federated queries in a distributed interlinked knowledge base.

Cheers,


Bob

Received on Tuesday, 28 September 2010 10:15:10 UTC