W3C home > Mailing lists > Public > public-owl-wg@w3.org > August 2008

Re: Support for OWL 2?

From: Rinke Hoekstra <hoekstra@uva.nl>
Date: Thu, 21 Aug 2008 02:55:22 -0700 (PDT)
Message-ID: <037ae6c0-34d3-49ad-93c7-ffadf0a72a77@x41g2000hsb.googlegroups.com>
To: TopBraid Composer Users <topbraid-composer-users@googlegroups.com>
Cc: public-owl-wg@w3.org

CC: to OWL WG

Hi Holger,

It would be good if you could make your/TopQuadrant's views known to
the OWL WG. Especially since a lot of effort is currently put in to
draft use-cases, any input that could either make the language appeal
to or cater for a wider audience is immensely valuable.

The OWL WG is trying its best to bring together a broad range of
users, and *specifically* not just the DL audience. Prime examples are
the OWL QL and RL profiles that have been designed to cater for
database-like applications. In particular the RL profile can be
regarded as a crossover between the DL and RDF world.

Thanks,

   Rinke


On Aug 21, 6:01am, Holger Knublauch <hol...@topquadrant.com> wrote:
> Hi Anita,
>
> we have basically frozen any further work on OWL 2 (formerly known as 
> OWL 1.1) until we have enough evidence of strong customer demand plus 
> sufficiently stable specifications. We are certainly monitoring what's 
> happening in this effort, but so far I am rather underwhelmed by the 
> outcomes of the OWL 2 working group, or in other words they do not 
> overlap with the needs of customers that we see in the real world. 
> Many of the OWL 2 features are only relevant for certain types of 
> tools (e.g. OWL DL inference engines) and we will leave these features 
> to those specialized tools and integrate them if available. Some OWL 2 
> features such as XSD facets (value constraints on properties) are 
> certainly relevant for TBC as an editing environment and we will 
> implement those things if there is enough interest from customers.
>
> If you (or others) could clarify which aspects of OWL 2 are especially 
> relevant for you this would be of great help.
>
> Regards,
> Holger
>
> On Aug 20, 2008, at 8:23 PM, Anita Kulkarni-Puranik wrote:
>
> > Since the OWL 2 specs are still in the recommendation - please 
> > could the Topbraid team let the users know the roadmap of support 
> > for OWL 2? specifically the support for operators in the value 
> > fields for properties?
>
> > thanks & regards,
>
> > Anita
>
> > On Tue, Jul 8, 2008 at 5:33 PM, topbraid-composer-users group <noreply@googlegroups.com
> > > wrote:
>
> > TopBraid Composer Users
> >http://groups.google.com/group/topbraid-composer-users?hl=en
>
> > topbraid-composer-users@googlegroups.com
>
> > Today's topics:
>
> > * recursion? - 4 messages, 4 authors
> > http://groups.google.com/group/topbraid-composer-users/browse_thread/...
> > * Extending SPARQLMotion - 2 messages, 2 authors
> > http://groups.google.com/group/topbraid-composer-users/browse_thread/...
>
> > =
> > =
> > =
> > =
> > =
> > =
> > =
> > =
> > ======================================================================
> > TOPIC: recursion?
> >http://groups.google.com/group/topbraid-composer-users/browse_thread/...
> > =
> > =
> > =
> > =
> > =
> > =
> > =
> > =
> > ======================================================================
>
> > == 1 of 4 ==
> > Date: Sun, Jul 6 2008 1:31 pm
> > From: "Johnrobert Gardner"
>
> > I realize I may be hoping ahead-of-features, or just failing to 
> > recognize
> > the obvious, but I've not figured a way to do recursion in SPARQL 
> > (wondering
> > if SPARQLMotion has this?) - reason being, I have to query, somewhat 
> > akin to
> > tops:, for parentage, currently doing so like this (it's a flat 
> > schema in
> > the source:, so no rdfs:subClassOf &c. to leverage):
>
> > ?fac a source:building
> > ?fac source:hasParent ?fac2
> > ?fac2 source:hasParent ?fac3
> > ?fac3 source:targetType 'foo'^^xsd:string
>
> > SOmetimes it's just up to fac2, sometimes fac4 or 5. I'd like to 
> > wirte one
> > script that recurses against any possible number of steps. Google 
> > seemed to
> > indicate any recursion in sparql is an extensions/will-add/future 
> > feature --
> > so wondered if I was either missing something and, similarly, 
> > missing such a
> > feature in sparqlmotion?
>
> > thanks,
>
> > == 2 of 4 ==
> > Date: Mon, Jul 7 2008 6:39 am
> > From: "Schmitz, Jeffrey A"
>
> > Hello,
> >  I'm not sure I understand exactly what you're looking for. If you
> > just want to recursively find all the parents of buildings (do 
> > buildings
> > have parents?) that are of targetType 'foo', as long as the parents of
> > buildings are also buildings (or can be inferred to be buildings by
> > using domain/range on the hasParent property), this will recursively 
> > do
> > it.
>
> > ?fac a source:building
> > ?fac source:hasParent ?facParent
> > ?facParent source:targetType 'foo'^^xsd:string
>
> > Your example will recursively find all "grandparents" of buildings 
> > that
> > have targetType 'foo', again, as long as all parents of buildings are
> > also buildings. You do have to pick an ancestry level that you want 
> > to
> > do the check for being targetType 'foo' (I don't think anything else
> > really makes sense anyway), but once you do that, sparql will work
> > recursively.
>
> > Does this answer your question?
>
> > Jeff
>
> > ________________________________
>
> > From: Johnrobert Gardner [mailto:johnrobert.gard...@gmail.com]
> > Sent: Sunday, July 06, 2008 3:32 PM
> > To: topbraid-composer-users
> > Subject: [tbc-users] recursion?
>
> > I realize I may be hoping ahead-of-features, or just failing to
> > recognize the obvious, but I've not figured a way to do recursion in
> > SPARQL (wondering if SPARQLMotion has this?) - reason being, I have to
> > query, somewhat akin to tops:, for parentage, currently doing so like
> > this (it's a flat schema in the source:, so no rdfs:subClassOf &c. to
> > leverage):
>
> > ?fac a source:building
> > ?fac source:hasParent ?fac2
> > ?fac2 source:hasParent ?fac3
> > ?fac3 source:targetType 'foo'^^xsd:string
>
> > SOmetimes it's just up to fac2, sometimes fac4 or 5. I'd like to 
> > wirte
> > one script that recurses against any possible number of steps. Google
> > seemed to indicate any recursion in sparql is an
> > extensions/will-add/future feature -- so wondered if I was either
> > missing something and, similarly, missing such a feature in
> > sparqlmotion?
>
> > thanks,
>
> > == 3 of 4 ==
> > Date: Mon, Jul 7 2008 11:17 am
> > From: Holger Knublauch
>
> > Hi JR,
>
> > are you only interested in getting the top-most hasParent value, or do
> > you want to get them all? In the latter case you could first run an
> > inference rule such as
>
> > CONSTRUCT {
> >    ?a source:hasParent ?c .
> > }
> > WHERE {
> >    ?a source:hasParent ?b .
> >    ?b source:hasParent ?c .
> > }
>
> > to "flatten" the hierarchy and then do the "real" query. You can
> > configure TopBraid inference engine so that the system would run those
> > rules directly before doing the query (most efficiently using Jena 
> > rule
> > syntax for the example above).
>
> > In order to get full-blown recursion, you can leverage user-defined
> > SPARQLMotion functions to some extent, because any user-defined 
> > function
> > could run itself again (make sure not to run into infinite loops, as 
> > the
> > cancel button may not work :) ). I am not convinced that you need 
> > this
> > for the above scenario though.
>
> > Holger
>
> > Johnrobert Gardner wrote:
> > > I realize I may be hoping ahead-of-features, or just failing to
> > > recognize the obvious, but I've not figured a way to do recursion in
> > > SPARQL (wondering if SPARQLMotion has this?) - reason being, I 
> > have to
> > > query, somewhat akin to tops:, for parentage, currently doing so 
> > like
> > > this (it's a flat schema in the source:, so no rdfs:subClassOf &c. 
> > to
> > > leverage):
>
> > > ?fac a source:building
> > > ?fac source:hasParent ?fac2
> > > ?fac2 source:hasParent ?fac3
> > > ?fac3 source:targetType 'foo'^^xsd:string
>
> > > SOmetimes it's just up to fac2, sometimes fac4 or 5. I'd like to 
> > wirte
> > > one script that recurses against any possible number of steps. 
> > Google
> > > seemed to indicate any recursion in sparql is an
> > > extensions/will-add/future feature -- so wondered if I was either
> > > missing something and, similarly, missing such a feature in 
> > sparqlmotion?
>
> > > thanks,
>
> > == 4 of 4 ==
> > Date: Mon, Jul 7 2008 11:47 am
> > From: "Jeremy Carroll"
>
> > Another thought:
>
> > Depending on your data, computing the transitive closure of hasParent,
> > following Holger's suggestion is a good solution, but if you have a 
> > lot of
> > hasParent links, then this might be quite a slow precomputation.
> > If
> > ?fac3 source:targetType 'foo'^^xsd:string
> > is selective (e.g. matches only one or two triples)
> > then having a custom solution that works backwards from 
> > 'foo'^^xsd:string
> > may well be the most effective.
>
> > I would follow Holger's suggestion, and only start looking at other
> > solutions if there are performance issues
>
> > Jeremy
>
> > -----Original Message-----
> > From: Holger Knublauch [mailto:hol...@topquadrant.com]
> > Sent: Monday, July 07, 2008 11:18 AM
> > To: topbraid-composer-users@googlegroups.com
> > Subject: [tbc-users] Re: recursion?
>
> > Hi JR,
>
> > are you only interested in getting the top-most hasParent value, or do
> > you want to get them all? In the latter case you could first run an
> > inference rule such as
>
> > CONSTRUCT {
> >    ?a source:hasParent ?c .
> > }
> > WHERE {
> >    ?a source:hasParent ?b .
> >    ?b source:hasParent ?c .
> > }
>
> > to "flatten" the hierarchy and then do the "real" query. You can
> > configure TopBraid inference engine so that the system would run those
> > rules directly before doing the query (most efficiently using Jena 
> > rule
> > syntax for the example above).
>
> > In order to get full-blown recursion, you can leverage user-defined
> > SPARQLMotion functions to some extent, because any user-defined 
> > function
> > could run itself again (make sure not to run into infinite loops, as 
> > the
> > cancel button may not work :) ). I am not convinced that you need 
> > this
> > for the above scenario though.
>
> > Holger
>
> > Johnrobert Gardner wrote:
> > > I realize I may be hoping ahead-of-features, or just failing to
> > > recognize the obvious, but I've not figured a way to do recursion in
> > > SPARQL (wondering if SPARQLMotion has this?) - reason being, I 
> > have to
> > > query, somewhat akin to tops:, for parentage, currently doing so 
> > like
> > > this (it's a flat schema in the source:, so no rdfs:subClassOf &c. 
> > to
> > > leverage):
>
> > > ?fac a source:building
> > > ?fac source:hasParent ?fac2
> > > ?fac2 source:hasParent ?fac3
> > > ?fac3 source:targetType 'foo'^^xsd:string
>
> > > SOmetimes it's just up to fac2, sometimes fac4 or 5. I'd like to 
> > wirte
> > > one script that recurses against any possible number of steps. 
> > Google
> > > seemed to indicate any recursion in sparql is an
> > > extensions/will-add/future feature -- so wondered if I was either
> > > missing something and, similarly, missing such a feature in 
> > sparqlmotion?
>
> ...
>
> read more 
Received on Thursday, 21 August 2008 09:55:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 21 August 2008 09:56:01 GMT