See also: IRC log
<tlr> http://www.w3.org/2004/12/rules-ws/paper/64
<tlrDC> Scribe: tlrDC
Adrian: Everything downloadable in machine-readable form?
Christine: http://idm.univ-rennes1.fr/~odameron/anatomy/abstractModel/index.html
Boley: functions?
... negation? ...
Christine: Don't want to go into that discussion.
Ian: "need rules" means "larger subset of FOL", regardless of the way in which it's written down.
... you need more expressive power, whether you get it through rules or first-order sentences may not matter ...
Christine: ??
Ian: If I give you full FOL, happy?
... from an expressivity point of view? ...
timbl: OWL-DL -- how much of OWL do you use?
christine: If you don't express something with OWL-DL, use rules instead?
Phil Archer: Talking about web sites. Porn. Gambling. Nazis. Everything else we don't want your children to see.
scribe: about ICRA ...
http://www.w3.org/2004/12/rules-ws/paper/18
Phil Archer: ... rating, self-regulation, have done that for the last 10 years, still generating PICS labels ...
scribe: move from PICS to RDF-based labels ...
... PICSrules for about 8 years ...
... web site doesn't fit with RDF model ...
... BBC site: 5e6 pages, don't want to put meta tags on all of these ...
... want one description ...
... need to be able to work with major companies' work flows ...
... idea to put PICS label into content doesn't fit with work flow ...
... no matter who creates content, point to where the rating is ...
... first use case for rules, URI -> description ...
... link rel tag, HTTP header that points to description's location ...
<DanC_lap> hmm... referring to a "block" of RDF
[[ Archer walks through an example of their current rules-in-RDF language. ]]
scribe: trust marks, as part of Quattro project ...
... these logos aren't machine-understandable ...
... trust marks as a use case ...
... mobile ok system ...
... parental control ...
... if there's a better way to do the labeling, please tell ...
... reducing data to binary ...
... various dimensions of rating ...
... Titanic example -- rating + detail ...
... in the end: does the child see the movie? ...
... RDF data + user profile ...
Adrian: How does the user profile get in there?
Archer: Talk to Microsoft.
... parent decides what the child sees ...
BenjaminG: Capture policies and data in RuleML -- merging policies, P3P ...
... want to merge rules, have some priorities ...
... total ordering ...
[[ a bit hard to understand ]]
Paul?, Fair Isaac: Good use case, where rules hit the common man.
scribe: overrides, rule management (??) ...
Archer: Hope it's scalable -- Yahoo has one person applying labels.
http://www.w3.org/2004/12/rules-ws/paper/46
Doug Clark, Metier: Project Mgmt industry; model some rules and knowledge.
scribe: concept of a project becoming important asset within organization ...
<sandro> (this session is being scribed off-line)
<sandro> oh. two scribes. ah.
<sandro> :-)
scribe: compound rules, rule of rules ...
... full expressivity, mathematical and logical ...
[[ flow-chart ]]
Marc?: What's meant by the rule target -- that term is being used in different ways.
Clark: The target the rule will be applied to.
Marc?: Different rules applied to different data.
??: What's a compound rule?
Clark: Rule made of other rules.
... question could be, "should I fly" ...
... rules about weather at destination, weather at home ...
<DanC_lap> (sounds like a compound rule is a rule that conjoins results from other rules)
Clark: weather at home one rule, weather at destination another rule ...
http://www.w3.org/2004/12/rules-ws/paper/74
scribe: cjm: issues and challenges ...
... binary predicates make things more difficult to construct and understand ...
... to move that to SWRL, need throw-away class ...
... lots of code to catch one higher-order relation ...
... 9 relatively simple rules -- 15 lines of pseudo-prolog -- turn into > 1000 lines of SWRL ...
... mismatch between declarative semantics and implementation ..
<sandro> "a real nightmare to debug"
scribe: input/output terms? ...
<sandro> They have implemented about half of the SWRL builtins, but they have to determine modality for Jess to handle it -- they require only one unbound varaible.
scribe: generation/assertion of new facts ...
... currently stepping outside SWRL semantics ...
... next set of issues, time ...
... decision from partial information ...
<sandro> cjm: We need assert & gensym [[ ie production rules ]]
scribe: need negation of failure ...
... could be as in N3 ...
... model moving objects ...
... time stamping ...
... both of incoming data and inferences ...
<timbl> he likes the system of explicit scope with log:includes
scribe: wish list ...
<bijan> DLR!
<timbl> He has ideas of how to comile higher-arity relations into binary relations.
scribe: user-understandable traces ...
... performance! ...
... uncertainty ...
Dave?: OWL ontologies as vocabulary, not processing semantics?
cjm: Using some OWL axioms.
HaroldB: top 10 nice, please go back to SWRL issues
...
cjm: Syntax is horrible.
HaroldB: Could have nice presentation syntax which is binary.
cjm: Would help!
Srinivas Krovvidy: ...
scribe: mortgage industry ...
... treat rules as data ...
... mortgage process flow ...
<scribe> Scribe: tlrDCDCgDC
<scribe> Scribe: tlrDC
<scribe> Scribe: tlrDC
<scribe> Meeting: W3C Workshop on Rule Languages for Interoperability [Day 2]
Ed (NIST): What do you mean by "expressional completeness"?
Srinivas: some constructs in legacy rule language.
... need to be able to define our building blocks, whatever the language constructs ...
Ed (NIST): Can you make that language available?
??Abdullah: Mentioned MISMO, rules based on object model. Is MISMO an object model?
Srinivas: MISMO based on XML schema. Is data model for specifying any kind of data ...
?Abdullah: Not an object model in the Java or C++ sense?
Said: Need for something more than XML Schema for MISMO?
Srinivas: Parts not adopted. (?)
??: WRT MISMO, are MISMO biz rules published? Rules that go beyond expressive power of Schema?
Srinivas: MISMO is data schema. Not rules.
Said: It's just a data model for interchange between players in mortgage industry.
http://www.w3.org/2004/12/rules-ws/paper/36
DaveReynolds: Focus on rule languages in the semantic web.
... use cases by proxy ...
... have had rules language for some years ...
... communicate use cases we have seen ...
... Jena ...
... wanted to use rules as infrastructure for RDF-S and parts of OWL-full ...
... estd user base 20k
... sources of insight -- own use, people we bump into, user list ...
... no evidence of people sharing rules ...
... expressivity: negation over extension ...
... implementing ontologies by way of rules ...
... RDF-S, e.g., -- people don't rely on model theory, but on rules ...
... avoiding lock-in as motivation for standardizating rule language ...
... validation use case ...
JosDeRoo: semi-positive?
DaveReynolds: negation only of extensional predicates.
<sandro> (which I think means: the givens, like log:notIncludes)
CSMA: Policy use case, interoperability requirements?
DaveReynolds: Access control within closed system.
<Zakim> tim2, you wanted to ask: Do you handle/need collections? With builtins? What needs for collections sets and/or lists?
<Zakim> timbl, you wanted to ask: To what triples do you compile down Nary -- list? Class with properties? If the latter how do you do builtibs?
[[ DaveReynolds talking faster than TimBL... ]]
StefanDecker: ?
[[ DaveReynolds losing scribe again. ]]
<sandro> StefanDecker: Why don't you see users sharing rules?
<scribe> Scribe: sandro
DaveReynolds: Network effect, maybe? The sharing hasn't gone very far up the S-curve
<tlrDC> Scribe: tlrDC
em: General questions on jena list?
DaveReynolds: why not -- but people tend not to answer some questions.
<sandro> DR: The dev mailing list has a lot of people asking question. Not a lot of people answering.