----------------------------------------- Eric Miller: ---------------------------------------------- How do we get the Web to full-scale adopt the Semantic Web the same way they adopted HTML validation. This good for businesses and KR people due to the global network effect of the Web. When I see XML I see data that could be bound to open world ontologies. Keep your eyes on the prize The problem is that the Semantic Web is fundamentally a specification language, people need to DO things with data. Businesses process - a process is a program,a workflow is like a programming language without iteration. Formalizing a business process is moving towards a full-scale programming language, rules are moving towards a programming language, a function that maps to either true or false. W3C does NOT want is a programming language like Prolog 2.0 for the Web that remains firmly in hands of academia while remaining useless to business.We also don't want business models to not be easily implemented formally. We don't want to keep using Perl for getting data off the Web (SparQL) If you define a low-level interoperability standard that does not have an eye towards extension and then years later hack together a formal semantics. Ian horrocks: Declarative KR (results) vs. programming language (execution process) {ex:d test:point ?x, ?x math:sin ?y} => {...} dynamic loading of new functions on the web defaults and NAF don't work in open web Do work with respect to given graph reusalbe rules can't have implicit scope of deafult log:notInclues is used. log:semantics truth value gaps truth value partiallyiyt negation-as-failure - expresseds local completeness Closed-World assumptions. Dangers: Prolog 2.0 for the Web that stays firmly in the hands of the academic community or those members of the academic community that currently participate in the SemWeb community. Want something that is actually used by LOTS of people, including businesses. Now, there's 3 parties at here: 1) There is the Semantic Web community, hackers in the Stallman sense, which wants the Web to do more neat stuff and probably could care less about formal semantics as long as it works on a Web scale. 2) There is the academic logic community that shows either foresight or desperation for funding by jumping on the SemWeb bandwagon, and would like to build some rational FOL++ layer on the Semantic Web. 3) There are business users, who want a standard to save money, exchange rules, etc. Most of their rules are pretty simple, and they aren't even really sold on the Sem Web yet. Yet they are sold on a number of things the SemWeb hackers and the academic community isn't - flow of control, calling services, uncertainity, provenance, and connecting both to local code and Web Services, etc. So, here's the two main issues: 1) People think we can do RDF/OWL SemWeb stuff and FOL at the same time in our Rule language. Yes, you can translate FOL->RDF, you can probably translate ANYTHING into RDF, but you might get nonsense. Translating from RDF or OWL to FOL is easy, it's been done, written down. I might be wrong, but translating FOL to RDF or OWL is HARD while maintaining some notion of truth, it's like the difference between falling down and falling up (Pat Hayes). This is a big red light problem, since the business world does not really use RDF/OWL widely, but uses something more or less like FOL to express their business rules. 2) Also, the business community wants to do things that are NOT purely declarative and miss the good-old-fashioned I'm running prolog on my local machine metaphor. Business community may not see too much relevant in all these FOL/SemWeb stuff to what they have, which are thousands of rules. I saw almost NONE of the candidate technologies propose these -> flow of control, uncertainity, provenance, and connecting both to local code and Web Services, etc. Harold mentioned that he could put arbitary functions in Rule ML, that's great - and Tim mentioned that "dynamic loading of new functions on the web" - and almsot nobody else did. 3) It's clear that the SemWeb (we all agree a good thing => businesses semantic rules) needs FOL (plus some stuff), business world wants FOL (and something simpler Rulelight, not negation)) and the SemWeb vision (plus some perhaps more stuff, so it's a marraige made in heaven. Can we build FOL++ on top of RDF/OWL cleanly, b\c everytime I try to translate FOL into RDF it's very unclear what's going on? -> maybe once we get FOL++ we need to nail that on the head. (Tim and Harold). Rule light (frames) and Rule Full (FOL++). Can we build the additional features (flow-of-control, uncertainity, provenance (??)) that business world wants on top of FOLSemWeb++, and what does this do to our logical models, and are these PERHAPS another level on TOP of the FOL++ level that we seem to currently be debating.