% notes from JC telecon 3/9/04 % by Benjamin Grosof participants: Mike Dean Benjamin Grosof Ian Horrocks Harold Boley Peter Patel-Schneider Sandro Hawke Gerd Wagner Pat Hayes (joined at 5pm) o benchmarking prospects need good problems need good dimensions to measure problems incl. for artificial generation generate artificial problems as a waystation Benj: wrt rules: can look at Prolog and RDBMS literature, their test suites - e.g., XSB, size of backend database, number of joins and how many bindings for each joined variable, complexity of queries Sandro: e.g., Mercury wrt Prolog appears to have a dozen or so Ian: there aren't good problems even for OWL alone with lots of instance data Mike: see www.daml.org/data o planning for May and for Sept. Mike: want another SWRL version with more on built-ins Benj: there will be new effort on rules toolset in DAML, stay tuned REWERSE has a working group specifically on typing: number I3 led by Jan Maluszynki and (?) Uwe Assmann (see http://www.rewerse.net) Ian: there's a lot of fragmentation within, as well as between, KnowledgeWeb and REWERSE -- you could spend your whole life traveling to the meetings Mike: want to finish builtins within next month; then should we get into nonmon? Benj: I think we should get started on at least NAF and nonmon before the PI Meeting - is better understood theoretically than full SWRL actually . Gerd: indeed SWRL proof theory not well understood Harold: also can do closed-world modules Gerd: adding nonmon to full SWRL will be tough Benj: another option for focus: extensions towards first-order and Lloyd-Topor - is more straightforward - but is less compelling wrt use cases - related to SCL and DRS - propose we do our own first draft, then look at how relates to SCL and DRS - Ian: SCL is a framework, more than is intended for efficient operations Harold: let's have some use cases too i.e., examples and implementation experience - this could be low hanging fruit - Benj: this would feed well into new DAML Rules toolset effort, and be useful to people working on services and security Mike: looking at CLIPS, translating SWRL into it Benj: good opportunity there to use the mappings from SweetJess, so let's coordinate on that Sandro: another desirable direction to do in parallel: non-XML RDF syntax, perhaps Prolog-y or N3-y or KIF-y. presentation syntax / human-oriented string syntax - Benj and Harold: yes, and nicely complements the tools/implementation direction overall goal: - be in a position to have good discussion of issues, plans for following months, info on tools/implementations and use-cases/examples/benchmarks Pat H.: wrt SCL: the concept and abstract syntax and model-theoretic semantics is pretty stable, need some hammering out of XML syntax; hope readable draft by perhaps end of month Benj: how to coordinate with SWRL; it would be nice to make it very simple to implement translation between SWRL and SCL. what issues are there wrt that? Pat: issues include the following. SCL has only very generic/plain variable bindings; can type a quantifier only with an atom not a type, is very vanilla; also there is some provision for defining types, some hooks for that Pat: CL is moving again wrt becoming an ISO standard. Harry Delugash (spelling?) is in charge of putting it through ISO machinery, which takes a while, like 3 years. SCL is officially ad hoc, not directly trying to put it thru ISO. Benj: would it make sense to do it thru another standards body, e.g., Oasis which is relatively low overhead? Pat: yes, perhaps; or W3C or OMG. Sandro: how about making it a W3C submission? would be good input for Rules WG there. Pat: sounds good, if someone can provide some guidance. would view it as a semantic reference language rather than specific to Rules. %%% agenda for next week: more on built-ins, Mike and Said and Pat will confer in prep.