Re: Thinking on the semantics of Named Graphs

Yuzhong Qu wrote:

>>>>>2. Within an accepted graph, there is a triple saying some 
>>>>>unaccepted graph is a truth, or one graph log:implies another 
>>>>>graph (or other properties, such as eg:premise and eg:conclusion, 
>>>>>are used) . How about the meaning of these constructs?
>>>>
>>>>
>>>>The  meanings, seems to me, ought to follow from the model 
>>>>theories. Im not sure what you mean by an unaccepted graph ( 
>>>>unasserted?) . If it is claimed to be true by an asserted graph, it 
>>>>IS an asserted graph, right? Asserted (by X)  = claimed (by X) to 
>>>>be true; so if that includes a claim that Y is true, then Y is 
>>>>thereby asserted.
>>>>
>>>
>>>no,
>>>
>>>g assertBy p .
>>>
>>>only entails I(g) when we trust p (for instance if we are p). This is
>>>the asymmetry of performatives ....
> 
> 
> In other words, suppose X (a named graph) is accepted by Alice, and the truth of Y (another named graph) is asserted within X, and Y is not in the list of graphs accepted by Alice, then the truth of Y is not guaranteed to be accepted by Alice.
> 
> If I understand it correctly, then the assertion of the truth of a graph is differentiated from other assertions. 
> 
> Does it?

I am not sure if I assert a triple

g assertedBy p .

then what I am saying is not that g is true but that p has asserted it.
If I am p then you may conclude that I think that g is true.
If you are p and you believe me then you think that g is true, and so 
you may conclude g.
If neither you or I are p, and you believe me, then you may conclude that

g assertedBy p .

So, I don't think this is particularly different from other triples. 
It's special features only come into play when the object is playing 
some role in the conversation.

> 
> 
>>>At the wedding the bride and groom say "I do" believing their love will
>>>outlive the universe, whereas those around mutter "it won't last three
>>>months"
> 
> 
> Funny example of social meaning issue.
> 
> The "I do" assertions are "believed" by the priest, but only the marriage asserted by the priest is accepted.

That's interesting ... the performative in a wedding needs three people 
to assert it to make it valid: the bride, the groom and the priest.

In a typical contract we may have two parties both of whom may make a 
performative which would be nullified if:
- the other party does not make their performative
- the contract is illegal

> 
> 
>>Yes, I spoke carelessly, sorry. I should have said that if X asserts 
>>that Y is true, then to the extent that X is committed to the 
>>assertion of Y's truth, X is also committed to the assertion of Y. 
>>Of course if I don't believe X to be reliable, then I may place no 
>>more faith in that assertion of Y than I do in anything else X 
>>asserts.
>>
>>This all makes sense only if there is a way for one graph to talk 
>>about the truth of another, which I don't think there is, strictly, 
>>unless we count owl:imports as a kind of remote assertion.
> 
> 
> Yes, we need  such a way.
> 

Why?
I am not convinced that we need such higher-order capability in the 
semantic web; or if we do that it needs to be fully integrated with the 
lower order capabilities of RDF and OWL

Jeremy

Received on Friday, 2 July 2004 04:33:39 UTC