F2F5 Summary

Summary of discussions and decisions taken

• New Document Overview provides overview and document roadmap; this document will be rec track and the editor(s) will be "W3C OWL Working Group"; goal is to avoid many of the confusions that surfaced at LC; roadmap will be added to the beginning of each of the other documents; this will respond to several LC comments, including 10, 42, 49, 56(partial), 29 (partial), 34a and 27.
• It was agreed that we would follow the proposal for dealing with the Full/DL naming issue (LC comments 28, 48, 52b), plus making the structural syntax strongly typed as per LC comment 58. In addition, we will also add RDF syntax for examples in Syntax, thus addressing LC comments 29 and 48.
• Discussion about disjointness of numeric datatypes was inconclusive, although leaning towards disjointness (LC comment 24). Everyone should consult with their organisations; we will have a formal vote at the teleconf of 11th March.
• Regarding OWL/RIF datatype alignment (LC comments 22 and 25), we decided that we would send an email stating our reasons for supporting the datatypes that we do support and offering to work with RIF to define a suitable set of "web compatible" datatypes that are supported by both OWL and RIF.
• Regarding RL datatype support (LC comments 20 and 43b), we decided that RL will support all datatypes except rational and real (due to lack of implementations), with the decision on supporting double and float deferred pending the decision on disjointness of numeric datatypes. Sandro will additionally talk to RIF to find out if they would need other datatypes to be excluded in order to facilitate RIF based implementations of RL.
• Regarding nary hooks and rationale (LC comment 46), will reply that WG ran out of time/energy and point to "hooks" and nary Note.
• Regarding RL semantics (LC comment 61), will fix up conformance so that all profiles can be interpreted with either semantics (ACTION-298).
• Regarding QL, we will add reflexive, irreflexive, & asymmetric to QL profile (LC comment 18 & 19) but we will *not* add sameAs (LC comments 59) -- instead we will add words from Uli explaining how implementations could be extended to support sameAs.
• We will change the encoding of the property chains to a single-triple encoding (LHS is the superproperty and RHS is the list of properties), introducing owl:propertyChainAxiom as the name for the relevant property (part of LC comment 35).
• Regarding OWL/XML format (LC comments 29 & 35), we will respond to both with short rationale for retaining it as a rec, and we will add some text to NF&R motivating having an XML syntax (LC Comment 35) -- see email from Bijan.
• Regarding schedule, WG agreed to use its very best endeavours to be finished by September, but will hang in till December if needed -- see revised timeline. We will ask for an extension to end of year.
• Regarding imports in RDF-based semantics (LC comment 53), we agreed that ongoing improvements to non-LC doc will address this; basically, imports is *not* a semantic operation, and RDF-based semantics will defer to Syntax for imports.
• Regarding GRDDL, if we can have multiple GRDDL transforms, and systems will behave reasonably with them, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), AND (3) download single xslt (if a good one is produced); ELSE: we'll do the on-line transform service only.