Day 2
Morning
Petr on SPARQL-DL: mentioned direct-type; need to figure out whether it does that automatically when asking for rdf:type, or whether you need to be specific.
Mike Dean on annotations: interesting web based hierarchy view using partOf; also uses other preferred terms, e.g. isa, branchof, continuouswith, etc. See RadLex viewer.
Michel on Presto: interesting, could replace NMRR for some ontology use cases.
Protege plugin for modularization: need to try this out, although it sounds like it can't cope with ontologies which mutually import each other. Tell Sven & Nanda.
Lee on instance data: good talk, lead to talk with Clayton Fink about distinctions between clean and dirty camps and use cases. Deborah talked of it as a kind of lint.
Ordinance Survey. Uses "Kanga" methodology for helping non-OWL types to create decent ontologies;
Bijan on Pronto: needs examples and use cases; probability isn't intuitive for many (and neither is DL!);
Ontology Based Data Access: maps instance data from databases to OWL(LiteA); plugin to Protege; works for distributed queries over multiple data sources?
Anatomy Lens (IBM?): way to use ontologies to hone in on pubmed articles; gives explanations; let Mathis know.
Afternoon
RBAC in OWL. Interesting dichotomy between formal approaches and more expressive yet ungrounded approaches. Looks to be quite separate to XACML -> OWL stuff. Weren't able to express policies in DL.
Peter on DL safe rules; syntax is heading towards RIF?, which doesn't have an RDF/XML serialization.
Bijan on easy keys and n-ary datatypes: easy keys are a restriction on IFP for datatype properties, which sound reasonable; n-ary datatypes allows us to reason algebraically, e.g. to define relations between classes of units to achieve unit conversion; these are more about precise algebraic relations between classes than about procedural numeric computation; all over reals?
blah
