Re: Self-descriptive assertions

On Mar 23, 2004, at 1:40 PM, Mark Baker wrote:

> On Tue, Mar 23, 2004 at 11:37:42AM -0500, Bijan Parsia wrote:
>>> ??  I'm only interested, right now, in knowing whether the 
>>> *publisher*
>>> of that information asserts it.
>>
>> And the right answer should be, IMHO, "It depends".  So I'm against a
>> blanket mandate, especially in the media type.
>>
>> Who counts as the publisher? The owner of the server?
>
> Yes - the authority of the URI.  I understand that isn't sufficient for
> all cases, but it seems a reasonable starting point to me.

Just a data point: It doesn't for me.

>>>> (And will this affect, oh, DAWG? I.e., if I want to use an
>>>> application/rdf+xml as a query "by example", I won't be able to
>>>> because
>>>> it's asserted? I.e., my query wants to be *is* this bit of RDF/XML
>>>> asserted by you.)
>>>
>>> I can't make sense of that, but it looks interesting 8-)  Can you
>>> elaborate please?
>>
>> A question generally isn't an assertion. If my question is basically 
>> an
>> RDF/XML document (e.g., "is this graph true") then, on your scenario,
>> if I am the publisher of my query, then it can't have normal query
>> semantics.
>
> Only if you use that media type.

Yes. That's my point. I want to use that media type. I don't want to 
tie propositional attitude to the format. I see no advantage to it. I 
see a lot of existing contrary use (i.e., I'll have to *take down* 
currently published RDF/XML). It's easy to screw up with high potential 
cost.

>  See below.
>
>> I understand now. Ok, it was responsive. What you didn't realize is
>> that this group has been moribund for a while. Perhaps this will 
>> revive
>> things.
>
> That is my hope, yes.

Ok.

>>> I respectfully suggest that
>>> your concern would be best directed at "other useful uses of RDF/XML
>>> documents", which I agree with.  But I don't see how my suggestion
>>> interferes with that in general, nor specifically for aggregation.
>>
>> So, there's the duel problem: How someone indicates that *they've*
>> asserted such and such rdf. And how someone indicates that they've
>> *not* asserted such and such RDF.
>
> Yes.
>
>> Inside RDF, the quoting mechanisms
>> are, in a world, lame. (Reification, but reification is lame. Or
>> literals, but that  doesn't seem to be to most people's taste.)
>
> Yup. 8-)
>
>> But let me back up: I don't see how media type does the job you want.
>> How does media type help acertain who's the publisher?
>
> As above, it doesn't, the URI does that.  But I agree that richer
> mechanisms will be required in the future.

My counterpoint would be is that if you're going to make this sort of 
move, you need to jump right to something rather richer than what 
you've thus far offered (i.e., I need to have some sort of non-asserted 
publishing). Now, you have offered a possible future way of handling 
this (some other media type), but 1) we need something like this *now*, 
if we're going to break application/rdf+xml for people and 2) we 
shouldn't break application/rdf+xml for people, or, at least, me.

>> Do we really
>> want to force publishers to assert all the RDF they publish? (I think,
>> not. This was discussed extensively, I believe, in my last call
>> comments.)
>
> No, but if they want to publish unasserted RDF, then a new media type
> could be created for that purpose.

First, they do so want.

Second, I don't understand why we should do this with media types. I 
feel somewhat dense, but media types just seem so wrong for this task. 
I.e., I would *never* look at the media type document for this. Maybe 
I'm just dense, but it's really not obvious to me.

Cheers,
Bijan Parsia.

Received on Thursday, 25 March 2004 02:46:09 UTC