UWA WG Teleconference
16 Aug 2007


See also: IRC log


Stephane, Rhys, Dave, Kangchan, jcantera




<steph> hola amigos

<scribe> ScribeNick: Dave

Approve last telecon for publication

No objections.

<jcantera> I'm trying to call in

Welcoming Ryan Sarver as an Invited Expert

Ryan couldn't make it today, so we will try again next week

Stephane is still waiting to hear back from sysreq on the media wiki installation that we hope to use for the location work.

REX PAG report and new work on Distributed DOMs

<Rhys> + 1 to not calling it REX

Dave explains about the REX PAG report and the switch of name to DDOM.

Dave invites volunteers for help with work on spec and requirements etc.

Discussion of Rhys's draft ontology

<Rhys> http://www.w3.org/2007/uwa/editors-drafts/DeliveryContextOntology/2007-08-31/DCOntology.html

Jose's feedback is at http://lists.w3.org/Archives/Public/public-uwa/2007Aug/0000.html

Rhys gives the pointer to his latest draft

and invites Jose to go through his feedback

Jose explains that the feedback is mainly from Diego Berrueta and Luis Polo from CTIC, who are both ontology experts.

The MyMobileWeb project is would like to develop a description of the delivery context and wants to be compatible with the W3C ontology.

Jose notes that he has been getting different results from different ontology tools.

Rhys comments that Protege isn't very good at determining which conformance level of the ontology is being used. Pellet is much better.

Rhys asks if he should add the ontology level to his document? Jose says yes.

Rhys thinks that we should trust Pellet in regards to testing the ontology confirms to OWL DL.

Jose concurs.

Jose comments about the disjoint mechanism used in the ontology and thinks we should use owl:disjointWith.

Rhys responds that this leads to a combinatorial explosion as the number of classes increase.

He adds that he followed the advice of the SemanticWeb best practices group

Jose asks Rhys to add an explanation of this to the the document.

Jose notes that Protege can help with the creation of the assertions, but Rhys notes that you would still get hundreds and hundreds of statements.

Owl 1.1 gives richer power and would be valuable to our work.

Rhys agrees to add an explanation of why the ontology isn't using disjointWith.

Jose then turns to his comment on names. He suggests that names or labels should be treated as annotations rather being part of the domain model.

Rhys asks Jose to send him more details on what Jose has in mind.

Jose agrees to send some examples.

Rhys explains that he is using see-also, but he would be happy to look at other approaches.

Rhys asks if Jose would be happy if we published a 1st draft with an explanation of where changes were expected to be made in future drafts.

Jose is okay with that.

Jose next talks about the Associated_UAProf_Entity Class.

Rhys says that the choice of not using the UAProf vocabulary is deliberate.

He had tried to capture the origins of properties (?)

but maybe that isn't worth including.

Jose then talks about units. This is an ongoing problem in the ontology community.

Rhys agrees that there is a problem and he sought advice from Tim Berners-Lee and Dan Connolly.

Jose explains his suggested approach.

Rhys explains the ambiguity that led him to the current approach.

He currently isn't covering instances in the ontology.

Jose mentions 2 approaches, one using schema subtypes, the other using rdf value.

Jose and Rhys discuss the idea of using blank nodes, but Rhys doesn't think that avoids the ambiguity issue.

Separately specifying the value and unit unfortunately allows for multiple conflicting assertions that result in ambiguity.

Rhys is interested in looking at schema subtypes as it moves the problem outside of RDF.

Rhys is torn between making the ontology very precise and making it scalable.

Rotan says the DDWG has asked for the ontology to include unit conversion factors.

Jose volunteers to do some further study of the problem.

Rhys notes that people like Dan Connolly wants to ensure that existing RDF reasoners can be used with the ontology. This will be an interesting discussion to have with the relevant groups. Jose agrees.

Jose isn't convinced about including conversion factors within the ontology.

Rhys responds that Rotan wanted this to ensure interoperability.

Rhys takes the point that doing so doesn't add anything in respect to reasoning with the ontology, it is rather a question of how the ontology is used.

Jose thinks we should avoid making assumptions about how the ontology is used.

We run out of time and Dave asks Rhys if he has enough to be getting on with.

Jose finally mentions the issue of representing specific device versus a class of devices.

Rhys notes that the existing ontology can represent both.