W3C home > Mailing lists > Public > semantic-web@w3.org > April 2015

Re: Question on dereferenceable RIF rules

From: Dave Reynolds <dave.e.reynolds@gmail.com>
Date: Wed, 08 Apr 2015 22:33:30 +0100
Message-ID: <55259EAA.1070307@gmail.com>
To: semantic-web@w3.org
Hi Michael,

A RIF rule shouldn't be an rdfs:Class, those are quite different 
concepts. But absolutely  anything is an rdfs:Resource, no need to 
declare it. The explicit declaration of the rdf:type of something isn't 
correlated to your ability to dereference it. So I don't think that need 
hold you up.

However, I'm not convinced that having a short cut for dereferencing a 
single rule as opposed to a rule set containing just one rule would make 
that much different.  In my experience rules typically come in sets 
designed to work together. Creating rules that can be usefully reused 
out of their original context is hard, the way URIs are assigned is the 
trivial part of the problem. That goes in spades for RIF-PRD but I 
assume you are here talking about declarative rules (RIF-Core or RIF-BLD).

I think there are lots of reasons RIF is a failure, and lots of history 
behind that, but I doubt that the lack of single rule import is really a 
significant part of that.

SWRL is a totally different beast ...

Best wishes,

On 07/04/15 09:43, Michael Petychakis wrote:
> Thank you very much Dave!
> Your answer was indeed helpful, and I was looking more into the details of the specification.
> What I would actually need in my case, would be to have a rule declared an rdfs:Resource or rdfs:Class. This way I could solve the issue of dereferencing and make it easier to  embed rules within my Linked Data documents.
> For example in OWL I can reference a specific class since those are rdfs:Class equivalent and a server can reference them by convention.
> Do you think that such an approach would actually make sense also in RIF(and/or SWRL)?
> This way I could declare that I have a rule, named Rule1 which is located in a remote server, reference it as
> http://example.org/mydocument#rule1
> and link it to another document, in a similar manner like the Linked Data paradigm.
> Of course this would need an updated reasoner implementation for RIF as proof of concept, but I think it could scale and make rules reusable over the web.
> At the moment I see almost no SWRL /RIF in any Linked Data documents around (based on my experience of course) and maybe there is a correlation with this.
> Kind regards,
> Michael Petychakis.
>> On Apr 6, 2015, at 8:12 PM, Dave Reynolds <dave.e.reynolds@gmail.com> wrote:
>> RIF rules can be identified by an IRI but the standards don't specify any dereferencing at an individual rule level.
>> The syntax is arcane, but the IRICONST? in the IRIMETA production [1] is an optional IRI to denote a rule (or indeed parts of rules).
>> You can, of course, pull in rule documents using the Import directive and so, as you say, dereference a document containing just one rule.
>> It would be perfectly reasonable to give the rule an IRI such as you suggest:
>>     http://example.org/mydocument#rule1
>> and then include that rule by importing the containing document. However, the standard doesn't enforce any relation between the document location and any IRI annotations on the contained rules. Nor does it require any dereferencing other than explicit Import.
>> This is no different from RDF and OWL. In the standards then RDF resources can be identified by IRIs (well, URI references) but there is no formal requirement to be able to deference them. In Linked Data we do expect to be able to deference them but that's a convention on top, not part of the RDF specification. Similarly with OWL there is an explicit document-level import but no axiom level dereferencing other than by convention.
>> Dave
>> [1] http://www.w3.org/TR/2013/REC-rif-core-20130205/#Annotations_and_Documents
>> On 06/04/15 14:55, Michael Petychakis wrote:
>>> Hello all,
>>> I am a little bit confused by reading the SWRL and RIF standards, and I am not even sure if this is the right list for this question.
>>> So, I would like to know if I can have a dereferenceable rule by a specific IRI.
>>> I could of course have only one rule declared within a referenced graph and then reference this graph, but I am looking if there is a simpler solution.
>>> Since both are web standards, there should be by default a way to reference a rule, for example something like:
>>> example.org/mydocument#rule1.
>>> If you have any ideas, reference, suggestions, they would all be highly appreciated.
>>> Thank you very much in advance,
>>> Michael Petychakis
>>> Electrical & Computer Engineer, Dipl Eng, Research Associate
>>> Decision Support Systems Laboratory
>>> School of Electrical and Computer Engineering
>>> National Technical University of Athens
>>> 9 Heroon Polytechneiou Str
>>> Zografou - Attica, 15773, Hellas
>>> Tel:  +30 210 7723555
>>> Fax: +30 210 7723550
Received on Wednesday, 8 April 2015 21:33:59 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:42:59 UTC