Re: RDF as a syntax for OWL (was Re: same-syntax extensions to RDF)

On Jan 7, 2005, at 12:35 AM, Jeen Broekstra wrote:

> Bijan Parsia wrote:
>
> [snip]
>
>> There isn't "a" problem. There's building ontologies in, tools for,
>>  applications using, and extensions of OWL (and RDF), plus teaching
>>  and explaining it to people. I've not found it useful for any of 
>> these, and typically not neutral for them either.
>
> My own experience in developing semantic web tools has not nearly been
> so heinous (although of course it wasn't always a pleasure either),
> which is probably why I have been questioning your finding.

You do know I'm talking about embedding more expressive logics in RDF? 
Right? My claims are limited to that.

I'm not a big fan of triple syntax more generally, but it's not 
*nearly* so bad.

So I fall into the camp of people who want to use RDF for "instance 
data" but want to extend its syntax as we add new semantic features.

> Also, in
> teaching about the semantic web stack I have not had any particular 
> problems with the fact that the knowledge is represented in graphs.

But that's a big underdescription of the situation. Did you cover RDF's 
model theory? The difference between OWLDL and OWL Full? The various 
correspondences with FOL? Did you cover proof procedures? How about 
layering?

>  The students did have problems with the RDF/XML syntax though.



> [snip]
>
>>> perhaps I have read that wrong however. If your only point is that 
>>> for some tasks (like nnf conversion and species validation),
>>>  triples are an unwieldy representation format, and that other 
>>> formats are much better suited for these tasks, then fine,
>> *Most* tasks. And dude, thus far I've come up with killer examples,
>>  and you've come up with nothing. Show me an implementation task 
>> that's only a *little* worse using the triple representation.
>
> There are tons of implementation tasks in which the triples do not
> "get in the way" and are even at times beneficial.

We are talking about the same-syntax embedding of OWL, right?

> Representing
> thesauri for example, or integrating multiple knowledge sources, or
> implementing semantic p2p systems,

Ah no.

> or querying RDF/RDFS/OWL,

Sound and complete querying? Esp. of OWL?

>  or...
> sheesh I don't know what type of example to quote. See my homepage[1]
> for some papers on application development with/for RDF/RDFS/OWL. See 
> the lists of applications that make use of Jena or Sesame or Redland, 
> SWI Prolog, or... The naked fact that production systems actually run 
> with this stuff should be evidence alone that it cannot be as much of 
> a development nightmare as you claim (and again: I'm not claiming it's 
> all wonderful shiny happy either).

Well, since the use of it to represent domain information isn't what's 
at issue, this is mostly irrelevant.

Furthermore, you just punt on the key point that I did not claim 
insurmountability. Production systems can compensate.

> Again: yes, for syntactic validation stuff like the examples you've
> quoted, I'm perfectly willing to accept that that is troublesome using
> triples (you've shown it, I believe it. We have no quarrel here). But
> for all your claims that this applies to "most tasks" where RDF and
> OWL are concerned, my own experience does not back this up. The
> problem may lie in that we may have very different ideas on what
> constitutes an RDF/OWL "task".

I'm talking pretty exclusively about things which are ultimately 
directed toward the use of OWL, especially as a natural fragment of 
first order logic. I, in general, also like to make conforming tools 
(to the specifications).

> That is all. I'm not claiming you are an idiot for saying the things
> you say, I'm not questioning your observation that for the tasks you
> tried to implement, triples were a burden. I'm trying to determine how
> much your claim can be generalized from your specific examples to your
> claimed "most tasks".

I also claim that triple syntax is less perspicuous for human beings 
(not just RDF/XML). I have only anecdotal evidence for that (see my 
mention of SWOOP earlier). I do hope to do a user study at some point.

>  I do admit that my insistence on using a
> different representation for the tasks you mentioned was based on a
> misunderstanding on my part, so sorry for that.

It's ok.

>  But I honestly don't
> understand why you find that so offensive, or why you insist on being
> personally insulting in return (in the same message in which you claim 
> to offer a 'group hug' no less).

Well, in your message, you go through and accuse me of defensiveness 
when I asked you to change your style, etc. etc. Only at the end do you 
offer an olive branch. I didn't feel like going back and rewriting all 
that, but I wanted to mark that I was overall at peace with your 
intentions up top. Guess that didn't work.

> Ah, let's just forget about it.

Sure thing.

> In any case, I'm bowing out of this conversation. My apologies to the 
> list for the scene all of this has caused (although I'm sure at least 
> some of you were mildly entertained).

Cheers,
Bijan Parsia.

Received on Thursday, 6 January 2005 16:37:00 UTC