Re: [UCR] ISSUE-12 and ACTION6198

Paul Vincent wrote:
> 1. For those of us who are outside of past W3C debates on the issue of "what is a standard W3C Semantic Web rule language", does anyone have any pointers / links as to why other attempts at this (RuleML? SWRL? Etc) have failed.

I think we are starting from different assumptions.

AFAIK there hasn't been any formal attempt to standardize a semantic web 
rule language, so no such attempts have yet failed, they just haven't 
happened yet.

A number of proposals for such languages have been developed including 
SWRL, SWSL, WRL and the REWERSE work. Those first three were all 
submitted to W3C [1][2][3] as input for some hoped-for standardization 
effort.

So when W3C came to charter a rule language working group within the 
semantic web activity some of the semantic web community expected it to 
be standardizing such a language starting from those formal submissions 
(plus practical experience with current in-use languages like CWM, Euler 
and maybe JenaRules).

Instead we have RIF. [*]

It is, I think, reasonably clear to much of the community that RIF is 
not that anticipated working group. This issue is about how we set 
expectations of what exactly the relationship *is* between RIF and such 
a language/language family.

For example, is some part of the working group going define a 
dialect/profile or two intended to be particularly suited to semantic 
web use? Are we going to take possible requirements for a future family 
of semantic web rule languages actively into account and do some 
foundational work beyond a minimal interpretation in the charter?

My interpretation of the discussion at F2F3 corresponds to "no" and 
"maybe" for those questions. I'm working on clarifying that "maybe".

> 2. For this subgoal (using any of the definitions mentioned in Christian's email), should there not be a subgroup of RIF for those who have particular interests with this, who can explain the requirements of a semantic web rule language where they differ from rule interchange? I suspect this subgoal has already been directing some of the reasoning of the members, and it would be healthy to be able to separate out these requirements so they can be costed and prioritised accordingly (eg where they extend the cost of rule interchange in an unreasonable way, or simplify rule interchange in a useful way). 

That sounds like a question for the chairs - I don't want to get stuck 
in process discussions.

> This subgroup could also explain how the semantic web + RIF-as-a-rule-language could support / represent the current use cases, which would be really useful to advance their case.

Several of the current use cases (7, 8, 10 and arguably 6) seem to me to 
already be directly motivated by using RIF in conjunction with the 
semantic web. Whilst the use cases involve rule interchange the rules 
that are interchanged in those cases are required to process RDF, 
interact well with OWL and so forth.

> 3. I support Christian's proposal, and would be interested to hear arguments against it.

See separate message.

Dave

[1] http://www.w3.org/Submission/SWRL/
[2] http://www.w3.org/Submission/SWSF-SWSL/
[3] http://www.w3.org/Submission/WRL/

[*] I have no insider knowledge but I assume that we have RIF rather 
than a rule language working group due to a mix of genuine interest from 
(at least) the business rules community in an interchange format and 
doubts as to whether there is yet sufficient consensus on the nature of 
a semantic web rule language to standardize one just now.

Received on Wednesday, 3 January 2007 12:38:24 UTC