Re: [UCR] ISSUE-12 and ACTION6198

Christian de Sainte Marie wrote:
> 
> Issue 12 [1] is the proposal to make an explicit goal of the WG that 
> "RIF should form a useful basis for a core semantic web rule language 
> (or small number of such)." (DaveR proposes: "RIF should be usable as 
> the basis for a Semantic Web rule language" [2]).
> 
> I have an action to propose text representing my position as a 
> counter-proposal (ACTION-198 [3]).
> 
> Here is the text I propose:
> ----------------------------------
> PROPOSED - The RIF WG will define a small number of standard dialects. 
> Although the design goal of each dialect will be rule interchange, any
> or all of these dialects may be considered standard semantic web rule 
> languages.
> ----------------------------------

As I tried to explain on the call yesterday this doesn't quite work for me.

It is certainly true that any RIF dialect could be thought of as a rule 
language - after all it will have a syntax and a semantics.

However the phrasing permits a reading where the WG might be doing 
something active to enable these dialects to be used as standard 
semantic web rule languages. This might accidentally raise expectations. 
Whereas my sense from the F2F3 discussion is this is unlikely and it 
would be better to say so.

The best alternative I've come up with so for is:

[[[
PROPOSED - The RIF WG will define a small number of standard dialects.
Although the design goal of each dialect will be rule interchange, each 
may be considered a rule language. Since RIF will support rules which 
can process RDF as data and will be compatible with OWL then any or all 
of these dialects could form the basis of some future standard semantic 
web rule languages. However, the RIF WG is not committed to developing 
any such proposals nor laying any particular foundations for them beyond 
the compatibility requirements mandated by the charter.
]]]

Given the F2F3 discussion I thought it more honest to replace "not 
committed to" by "will not" but that got a negative reaction too, hence 
the above.

---

This all seems a bit abstract. It might help to have some grounding. 
What would you want in a rule language before some reasonable developer 
might say "that's a semantic web rule language" rather than "that's just 
another RIF dialect"?

I'd suggest that might include (no claims of completeness here):

  a. the ability to infer not just process RDF assertions (which in turn 
raises the bNode-in-the-consequent issues)
  b. a clear picture on how to use OWL and the rule dialect together
  c. a set of function/builtins suited to the manipulation of RDF data 
(e.g. isURIRef, access to SPARQL data sources etc).
  d. a friendly syntax
  ...

So some of these we will do (for example (b) is mandated by charter), 
and some are out of scope (d :-) ).

Let us pick one of the boundary cases to ground the discussion. What 
about the set of builtins/functions such as one for access to an 
external SPARQL data source.

Technically there is nothing stopping us defining such a thing but where 
would that go? It can't go in RIF Core [*] because lots of rule vendors 
won't want to support such a thing. Unless we start a 
semantic-web-friendly dialect there is no dialect to put it in. We could 
put it in the proposed library of reusable components so that future 
definers of semantic-web-friendly dialects might reuse the same one. 
What I was trying to get at with my original proposal at F2F3 was how 
much of that sort of laying the foundations would the WG commit to. The 
above is my best shot at writing down the answer that I heard (as 
opposed to the answer that I might have preferred.)

Dave

[*] This statement may be too strong depending on the outcome of the 
discussion on what conformance to RIF Core means.

Received on Wednesday, 3 January 2007 18:45:47 UTC