Re: Alternatives to OWL for linked data?

Axel,

you might want to have a look at the profiles of OWL 2:

http://www.w3.org/TR/owl2-profiles/

which tries to strike a balance. The idea for all these profiles is that
you trade in expressiveness for an ease of implementation. Ie, I would
not write off OWL just yet!

You were asking about description logic programming; well, OWL 2 RL:

http://www.w3.org/TR/owl2-profiles/#OWL_2_RL

is exactly that: it is a manifestation of DLP. It has a Direct Semantics
'side', compatible with OWL 2 DL, and a rule based 'side', described by
the rule set:

http://www.w3.org/TR/owl2-profiles/#Reasoning_in_OWL_2_RL_and_RDF_Graphs_using_Rules

This rule set can be used for a forward or backward chaining approach
(or a combination thereof) that you describe. I have heard rumours
and/or statements on implementations coming up from various vendors. I
have, actually, a purely proof-of-concept-stupid-simple implementation
doing brute force forward chaining:

http://www.ivan-herman.net/Misc/2008/owlrl/

Just to show what happens. And I am sure other implementations will come
to the fore that I do not yet about.

The other profile that might become interesting is OWL 2 QL:

http://www.w3.org/TR/owl2-profiles/#OWL_2_QL

that is, in many respect, simpler than RL and may be very advantageous
when mapped on top of SQL. Again, I have heard rumours of
implementations for that profile, too.

We can also expect that both these profiles (and others) will play well
with the new version of SPARQL (well, actually, with the current one,
too). Because OWL 2 RL provides a clear deductive closure via the rules,
that one is certainly not a problem; I am not 100% sure about QL (but I
am sure there will be ways).

Cheers

Ivan

P.S. B.t.w., OWL 2 RL is also 'expressed' via RIF rules

http://www.w3.org/2005/rules/wiki/OWLRL

in case you have looked at RIF before


Axel Rauschmayer wrote:
> I'm currently reading Hendler's brilliant book "Semantic Web for the
> Working Ontologist". It really drove home the point that OWL is not a
> good fit when using RDF for *data* (names are generally not unique, open
> world assumption, ...).
> 
> But what is the alternative? For my applications, I have the following
> requirements:
> 
> - Properties: transitivity, inverse, sub-properties.
> - Resources, classes: equivalence. For my purposes, equivalence is a way
> of implementing the topic merging in topic maps [1].
> - Constraints for integrity checking.
> - Schema declaration: partially overlaps with constraints, serves for
> documentation and for providing default values for properties.
> - Computed property values: for example, one property value being the
> concatenation of two other property values etc.
> 
> The difficulty seems to me to find something universal that fulfills
> these requirements and is still easy to understand. Inference, when used
> for transitivity and equivalence, is simple, but when it comes to
> editing RDF, they can confound the user: Why can some triples be
> replaced, others not? Why do I have to replace the triples of a
> different instance if I want to replace the triples in my instance?
> 
> While it's not necessarily easier to understand for end users, I've
> always found Prolog easy to understand, where OWL is more of a challenge.
> 
> So what solutions are out there? I would prefer description logic
> programming to OWL. Does Prolog-like backward-chaining make sense for
> RDF? If so, how would it be combined with SPARQL; or would it replace
> it? Or maybe something frame-based?
>
> Am I making sense? I would appreciate any pointers, hints and insights.
> 
> Axel
> 
> [1] http://www.topicmaps.org/xtm/index.html#desc-merging
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf