RIF Working Group in a Nutshell

Dear All,

During the August 1st telecon I promised to provide some feedback to the
group regarding the RIF Wroking Group (WG). Since it seems that I will also
miss some of the next telecons (both on the 19th and the 3rd it seems that I
will be travelling) I am sending a mail with some feedback. 

==ID==
- RIF stands for Rule Interchange Format. 
- It is a Working Group (unlike our Interest Group), i.e. it will provide a
standard (W3C Recommendation).
- RIF is chartered for 2 years.
- RIF home page: http://www.w3.org/2005/rules/
- RIF Charter: http://www.w3.org/2005/rules/wg/charter.html 
- RIF wiki page: http://www.w3.org/2005/rules/wg/wiki/FrontPage

== Objectives ==
RIF is the current action of W3C for the rules layer of Semantic Web. The
primary purpose is to propose a format for interchanging rules between rule
systems on the (Semantic) Web and *NOT* to create a W3C Rule Language for
the Semantic Web. Thus, one should not expect to use RIF to represent its
rules, i.e. creating a RIF Rule Base. 

RIF is also expected to provide compatibility with current W3C Semantic Web
standards like RDF and OWL.

== Architecture ==
The work is split in to 2 phases:
In phase 1 we will defined a RIF Core language which is lets say "a minimum
*interesting* fragment that is common over most logic programming languages
and systems". Then in phase 2 several RIF Dialects will extend or restrict
the semantics and functionality of RIF Core elements to create a rule
interchange format for an LP language not supported by RIF Core. Examples of
RIF Dialects could be an F-Logic Dialect, a DisjunctiveDatalog Dialect, a
Horn+Negation, a Production Rule dialect, etc.

So if you wanted to exchange rules between your Disjunctive Datalog system
and some other one (also a Disjunctive Datalog system) you would have to
implement a mapping from your rule base to the appropriate RIF Dialect and
of course the other part should also be able to translate the RIF Rules to
its own format. 

Maybe the group will examine some cases of exchanging rules between diverse
systems, i.e. a RIF StableNegationToWellFoundedNegation Dialect but is only
expected to do it for cases that such mapping has been studied in the
literature and not do research on its own.

Checkout: http://www.w3.org/2005/rules/wg/wiki/Arch/Extensibility

== So Far ==
The group started with UseCases & Requirements and until then it has
produced 3 versions of them. The working version is here
http://www.w3.org/2005/rules/wg/wiki/UCR.

Currently the work is focused on RIF Core http://www.w3.org/TR/rif-core/,
its XML syntax, RDF Compatibility and build-ins. Originally RIF Core was
proposed to be Horn Logic + Sorts, but we have backtracked to Horn
http://www.w3.org/2005/rules/wg/wiki/Core.

Regarding dialects there was some work on a NegationDialect and some
Prodcution Rule System dialect but I think little has been done on this
issue. 

== RIF and Uncertainty ==
Since the beginning we (the NTUA-IVML group) have tried to bring up the
issue of uncertainty extensions of Semantic Web standards whenever possible.


Initially, we succeeded in having "uncertainty" mentioned in the RIF Charter
under the RIF Extensibility section
http://www.w3.org/2005/rules/wg/charter#extensibility.

Then during the UseCases work we added a UC for Fuzzy
http://www.w3.org/2005/rules/wg/wiki/Fuzzy_Reasoning_with_Brain_Anatomical_S
tructures but it didn't made it into the UCs document. Nevertheless, we
tried to add it implicitly through the Medical Decision support Use Case
http://www.w3.org/2005/rules/wg/wiki/UCR/Ruleset_Integration_for_Medical_Dec
ision_Support that exists in it.

As with the UseCases it seems that we could propose an Uncertainty Dialect
to the RIF group. Actually together with Carlos Damasio, Jeff Pan and
Umberto Straccia we have already done some work for an Uncertainty RIF
dialect in a paper to appear in Fundamenta Informaticae. 

But on the other hand from my experience in related W3C activities I would
say that it is unlike that such an extension will survive as a dialect in
the final standard. Uncertainty always looks kind of exotic to most people,
while several ones in RIF would definitely eagerly object having them as a
RIF Dialect. On the other hand even neutral people would prefer to see a
dialect covering their favourite or some popular system rather than
uncertainty if it gets to choosing among "n" for standardization. Moreover,
as far as I know, the issue of uncertainty LP is not such mature in the
sense that there are not so many uncertainty rule bases and systems out
there that one would like to interchange between them. So our case is
generally weak.

Currently, it is not decided how many and which dialects would the group
create as well as the requirements for Phase 2
http://www.w3.org/2005/rules/wg/wiki/Design_Constraints, where uncertainty
is included, have not been discussed yet.

Hope I was concise enough. I also welcome any related questions.

Greetings,
-gstoil

Received on Saturday, 8 September 2007 06:23:58 UTC