Re: [UCR] use case response

Jim Hendler wrote:

> At 18:36 +0000 2/21/06, Dave Reynolds wrote:
> 
>> Jim Hendler wrote:
>>
>>>  At 10:00 +0000 2/21/06, Dave Reynolds wrote:
>>
>>
>>>   >Now the WG could decide that this desire for a side-effect is in 
>>> conflict with the main goal, that the group should only be about rule 
>>> exchange and have no regard to the possibility of a semantic web rule 
>>> language. That would be just fine and would simplify my life 
>>> considerably.
>>>   >
>>>  I don't understand, the notion of a Sem Web rule language is okay 
>>> with me, but that isn't the same as what you said above
>>
>>
>> In that case I'm either not being clear (as usual) or 
>> mis-understanding what you mean by sem web rule language or, probably, 
>> both.
>>
>> I'm not sure I see the difference between what I'm asking for and the 
>> first use case in section 1.1 of the WG charter.
> 
> 
> I read use case 1.1 of the Charter to be primarily concerned with using 
> rules to map between schemas for different DBs (which I assume would be 
> expressed in RDFS or OWL) - I may have read too much into it.

I read it that way too.

That use case comprises:
   - someone integrating data sets from across the web
   - they write rules on a local system to perform the required mappings
   - they publish the resulting rules so that others can reuse them

I assume that the data is (or could be) exposed as RDF with structure 
described in RDFS/OWL.

To me this is a core use case for rules on the semantic web. It is of 
significant interest to us. The message transformation use case could be 
seen a special case of it in which only one source is mapped at a time.

Yet, despite it being in the charter document, there seems to be some doubt 
on whether that sort of problem is really a RIF use case.

It is true that the interchange of rules is not central to the use case. 
The interaction between rules and web information sources is central. The 
ability to publish the rules for potential future reuse is also central.

Your comments, Paul's and others seem to be setting a high threshold on 
there being in-your-face interchange going on for something to count as a 
RIF use case. Just publishing your rules so they could be reused by others 
doesn't seem to cross that threshold.

It would be good to have some use case like the charter one reflected more 
clearly in the UCR document (presumably in the information integration 
section that you are objecting to) or have it be ruled out of scope clearly 
enough that we don't have to waste more time probing this boundary.

>>>  I was primarily using "web rule" to mean a syntactic standard for 
>>> representing rules in one of the major Web languages  and explicitly 
>>> grounded in URI space - I generally use "Web ?x" to mean ?x being 
>>> webized as in [1]
>>
>>
>> So a rule language, with an XML serialization, in which all predicate 
>> symbols were URIs, which runs over data expressed in RDF (so variables 
>> get bound to RDF resources) and which can conclude RDF statements ... 
>> would that automatically be classifiable as a "web rule" language or 
>> would there need to be some deeper webization to qualify?
> 
> 
> I'd count that (and, in fact, at one point in the pre-history of this WG 
> had argued that it might be better to get something like that out first 
> and worry about these semantic things after).   That doesn't mean it 
> would be my first choice at this point, but if the WG decided that the 
> best way to address the charter was by creating something like that I 
> wouldn't see that we had been unresponsive to the charter.  That said, 
> you know quite well that I'm a big RDF fan, put a lot of my own skin in 
> the game to make sure OWL lived in the RDF world, and think it would be 
> foolish if we didn't at least have an RDF compatible syntax for the 
> rules (for example, a mapping via GRDDL from an XML schema would be 
> something I'd consider worth exploring if the WG was going in that 
> direction)

Sure, I wasn't meaning to exclude the option of an RDF compatible syntax, 
far from it.

Dave

Received on Thursday, 23 February 2006 17:21:59 UTC