Warning:
This wiki has been archived and is now read-only.
Chatlog 2009-01-14
From RIF
See original RRSAgent log and preview nicely formatted version.
Please justify/explain all edits to this page, in your "edit summary" text.
(Some of the original chatlog carries over onto the next day because that log uses UTC timestamps.)
Day 1 of F2F12.
00:00:00 <sandro> PRESENT: csma, gary, adrian, harold, kifer, jos, sandro, paul, changhai 00:00:00 <sandro> REMOTE: chrisw, dave, stella, leora, mikedean, hassan <sandro> Chair: csma, sandro 09:05:43 <RRSAgent> logging to http://www.w3.org/2009/01/14-rif-irc 09:06:03 <sandro> zakim, who is here? <sandro> Topic: Welcome, Logisitics, Agenda 09:06:06 <Zakim> apparently SW_RIF()11:00AM has ended, sandro 09:06:09 <Zakim> On IRC I see RRSAgent, AdrianP, sandro, mdean, Hassan, trackbot, Zakim 09:06:16 <sandro> zakim, this will be rif 09:06:22 <Zakim> ok, sandro, I see SW_RIF(F2F12)11:00AM already started 09:07:17 <Zakim> +Anish_Karmarkar 09:07:47 <sandro> zakim, Anish_Karmarkar is RIF_Meeting_Room 09:07:47 <Zakim> +RIF_Meeting_Room; got it 09:10:30 <sandro> scribenick: Harold 09:09:02 <sandro> zakim, who is here? 09:09:02 <Zakim> On the phone I see Mike_Dean, Hassan_Ait-Kaci (muted), RIF_Meeting_Room 09:09:06 <Zakim> On IRC I see cke, PaulVincent, Harold, csma, RRSAgent, AdrianP, sandro, mdean, Hassan, trackbot, Zakim 09:13:53 <sandro> Zakim, RIF_Meeting_Room has Christian, Gary, Harold, Adrian, Jos, Paul, Micheal_Kifer, Changhai, Sandro 09:13:53 <Zakim> +Christian, Gary, Harold, Adrian, Jos, Paul, Micheal_Kifer, Changhai, Sandro; got it 09:15:18 <sandro> zakim, who is on the phone? 09:15:18 <Zakim> On the phone I see Mike_Dean, Hassan_Ait-Kaci (muted), RIF_Meeting_Room 09:15:21 <Zakim> RIF_Meeting_Room has Christian, Gary, Harold, Adrian, Jos, Paul, Micheal_Kifer, Changhai, Sandro 09:22:56 <Harold> Agenda amendments: 09:23:16 <Harold> * One hour on safeness 09:24:19 <Harold> * Harold's PRD read 09:26:54 <Harold> Test 09:26:57 <Harold> Cases 09:27:44 <PaulVincent> Question on Conflict Resolution part in PRD agenda: does this not also apply in some way to logic languages (eg in Prolog, selection is order related)...? Just in case this needs to be specified in Core even if there is a simple default for BLD et al... 09:27:45 <Zakim> +[IBM] <sandro> Topic: Test Cases <sandro> Summary: Various RDF-related test cases were discussed <sandro> SubTopic: RDF_Combination_Blank_Node 09:28:07 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Blank_Node 09:28:33 <sandro> Zakim, IBM is temporarily Stella 09:28:33 <Zakim> +Stella; got it 09:28:39 <StellaMitchell> zakim, ibm is temporarily me 09:28:39 <Zakim> sorry, StellaMitchell, I do not recognize a party named 'ibm' 09:31:11 <Harold> Christian: Can you derive this in BLD, Core, ...? 09:31:51 <Harold> Jos?Michael: Yes, in BLD. 09:32:10 <Harold> s/Jos?Michael/Jos+Michael/ 09:34:13 <sandro> csma: if there were a named node, I'm sure it would work. I'm not sure about the bnode 09:35:24 <sandro> sandro: typical RDF forward-chaining system just do this by inventing names, and marking them as invented. 09:37:25 <Zakim> + +44.145.441.aaaa 09:37:30 <sandro> welcome, Dave. We're talking about http://www.w3.org/2005/rules/wiki/RDF_Combination_Blank_Node 09:39:10 <PaulVincent> Issue: how blank RDF nodes map to PRD (eg object-based rule engines)... probably it is a local variable, so should be no issue... 09:39:10 <trackbot> Created ISSUE-89 - How blank RDF nodes map to PRD (eg object-based rule engines)... probably it is a local variable, so should be no issue... ; please complete additional details at http://www.w3.org/2005/rules/wg/track/issues/89/edit . 09:39:16 <PaulVincent> Oops 09:42:22 <sandro> Stella: is this Core or SafeCore? 09:42:50 <sandro> Jos: this is safe 09:43:07 <DaveReynolds> Surely it is not safe, we don't support exists in conclusions 09:43:37 <sandro> PROPOSED: approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Blank_Node for BLD, Core, PRD, and Safe-Core 09:44:22 <Harold> Stella: Core sect 5.1? 09:44:25 <sandro> http://www.w3.org/2005/rules/wiki/Core#Safeness 09:44:44 <sandro> Jos: I believe this "existential" part is an error 09:44:49 <GaryHallmark> dave, a test conclusion is actually a rule condition, not a rule conclusion 09:45:56 <DaveReynolds> object 09:48:04 <sandro> forall ?X if ?X["http://www.w3.org/1999/02/22-rdf-syntax-ns#type"^^rif:iri -> <http://example.org/example#named>] then ex:a[ex:b -> ex:c] 09:48:22 <sandro> test case conclusion: ex:a[ex:b -> ex:c] 09:49:31 <sandro> forall ?X if ?X["http://www.w3.org/1999/02/22-rdf-syntax-ns#type"^^rif:iri -> <http://example.org/example#named>] then test:passed(test:RDF_combination_Blank_node) 09:50:15 <sandro> DaveReynolds: Either is fine. 09:50:35 <Harold> Dave: Withdraw objection. 09:51:30 <sandro> PROPOSED: approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Blank_Node for BLD, Core, PRD, and Safe-Core 09:51:39 <sandro> +1 09:51:45 <GaryHallmark> +1 09:51:54 <AdrianP> +1 09:51:57 <Harold> +1 09:52:00 <Hassan> +1 09:52:01 <josb> +1 09:52:02 <sandro> RESOLVED: approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Blank_Node for BLD, Core, PRD, and Safe-Core 09:52:07 <Michael_Kifer> +1 <sandro> SubTopic: RDF_Combination_Constant_Equivalence_1 09:52:29 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_1 09:55:42 <sandro> Sandro: this is basically a demonstration of what rif:iri means. 09:56:02 <DaveReynolds> But rif:iri is not a datatype in the RDF sense, it is a symbol space. 09:58:28 <sandro> jos: Indeed it's not... 10:02:34 <sandro> Sandro: If we approve this, I'm likely to use it as a hammer to to say "rif:iri" really is a datatype... 10:02:53 <sandro> Jos: We could say rif:iri as a dt isn't defined in these combinations... 10:03:19 <sandro> Sandro: I think we'd better got with "not defined"... 10:04:23 <josb> http://www.w3.org/TR/rif-rdf-owl/#Common_RIF-RDF_Interpretations 10:05:12 <sandro> jos: Condition 6 implies this... 10:06:35 <sandro> Issue: Can rif:iri be used as a datatype in RDF graphs, in combination with RIF? ie approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_1 10:06:36 <trackbot> Created ISSUE-90 - Can rif:iri be used as a datatype in RDF graphs, in combination with RIF? ie approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_1 ; please complete additional details at http://www.w3.org/2005/rules/wg/track/issues/90/edit . 10:07:23 <DaveReynolds> Another way of putting it: is rif:iri *required* to be in all datatype maps. 10:07:36 <Harold> Stella: Do we need XML version for all test cases? 10:07:43 <Harold> Sandro: yes. 10:07:56 <Harold> ... no preferred one. 10:08:09 <Harold> ... depends on platform. 10:08:24 <StellaMitchell> for the imported RDF documents, that is 10:10:00 <DaveReynolds> Please could someone post the link to the Test Case you are talking about? <sandro> SubTopic: RDF_Combination_Constant_Equivalence_2 10:10:04 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 10:11:56 <sandro> DaveReynolds: Surely this requires d-entailment? 10:12:08 <sandro> Jos: Nope, even with Simple you get RIF datatype support. 10:14:37 <sandro> Sandro: There's no way to get RDF out of RIF. There's no export. So xs:string deosn't matter. 10:16:08 <Harold> Christian: explanins RDF-RIF interchange on the whiteboard. 10:16:52 <sandro> Jos: Who cares. Yes, when you pass the rdf-manipulating-rule through RIF, you may find one is producting "foo" and the other "foo"^^xs:string. No one really cares about the difference. 10:17:57 <sandro> csma; round tripping could give you both 10:18:07 <sandro> Jos: this is like preserving the order of triples 10:19:29 <DaveReynolds> I agree the test case is fine. Jos which is condition in the document that specifies that mapping for plain literals? I can find the informative section and the formal conditions for well-typed literals but can't immediately spot the formal condition that covers plain literals. 10:21:47 <sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 (for all dialects, and it's Safe) 10:21:52 <josb> The RDF semantics says that plain literals are mapped to strings 10:21:54 <josb> http://www.w3.org/TR/rdf-mt/#interp 10:21:57 <AdrianP> +1 10:22:06 <sandro> +1 10:22:11 <josb> +1 10:22:12 <Hassan> +1 10:22:14 <DaveReynolds> +1 10:22:14 <Harold> +1 10:22:17 <mdean> +1 10:22:17 <GaryHallmark> +1 10:22:20 <josb> condition 6 in http://www.w3.org/TR/rif-rdf-owl/#Common_RIF-RDF_Interpretations 10:22:23 <Michael_Kifer> +1 10:22:28 <sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 (for all dialects, and it's Safe) <sandro> SubTopic: RDF_Combination_Constant_Equivalence_3 10:22:48 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_3 10:23:13 <josb> says that typed literals in the RDF part are interpreted as the same object as their counterpart in RIF 10:24:12 <Zakim> + +1.212.781.aabb 10:24:24 <LeoraMorgenstern> zakim, aabb is me 10:24:24 <Zakim> +LeoraMorgenstern; got it 10:24:30 <sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_3 (for all dialects, and it's Safe) 10:24:43 <AdrianP> +1 10:24:48 <josb> so, the plain literal "a" in RDF is mapped to the character sequence "a" 10:24:52 <Harold> +1 10:24:53 <sandro> +1 10:24:56 <Hassan> +1 10:24:58 <GaryHallmark> +1 10:25:03 <josb> in RIF, the symbol "a"^^string is mapped to "a" 10:25:04 <sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_3 (for all dialects, and it's Safe) 10:25:14 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_1 10:26:17 <Harold> Jos: RDF graphs in the conclusion, instead of in the condition. 10:27:03 <josb> by condition 6, the typed literal "a"^^string is interpreted in the same way as the RIF constant "a"^^string 10:27:39 <sandro> postpost http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_1 related to issue-90 <sandro> SubTopic: RDF_Combination_Constant_Equivalence_Graph_Entailment_2 10:28:50 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_2 10:30:44 <josb> http://www.w3.org/TR/rif-rdf-owl/#Entailment 10:30:57 <sandro> Sandro: I am very confused -- how come the test case conclusion is in Turtle, not RIF PS ? 10:31:47 <sandro> Jos: This would be true in RDF only with D-entailment and the dt xs:string. 10:32:41 <Harold> Jos: Used Turtle syntax for conclusions. 10:32:58 <Harold> ... to show properties of the language. 10:34:04 <josb> see http://www.w3.org/TR/rif-rdf-owl/#Embedding_Simple_Entailment for implementation of this kind of thing 10:35:17 <Harold> Jos: if you use simple entailment, this is not an entailment. 10:35:33 <StellaMitchell> the ImportSupport property is supposed to help implemantation select which tests are applicable to them 10:35:38 <Harold> Sandro: concerned that we use different languages. 10:36:01 <Harold> Jos: all based on examples from document. 10:38:02 <Harold> Sandro: Perhaps import Turtle and use RIF for rule/conclusion. 10:39:42 <sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_2 but be sure to make it clear that this is a different kind of test case, about RDF entailment. (Current wiki template does not properly show that.) 10:40:07 <StellaMitchell> sandro, wiki template shows ImportSupport 10:40:38 <Harold> Harold: slot of different formats (RIF, Turtle) from the Test Case Document should be rendered to illustrate interop. 10:41:13 <josb> +1 10:41:18 <Harold> +1 10:41:22 <sandro> +1 10:41:22 <Hassan> +1 10:41:32 <GaryHallmark> +1 10:41:35 <AdrianP> +1 10:41:36 <DaveReynolds> 0 10:41:37 <sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_2 but be sure to make it clear that this is a different kind of test case, about RDF entailment. (Current wiki template does not properly show that.) 10:41:46 <sandro> (For all languages) 10:42:02 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 10:42:52 <sandro> ACTION: jos to add a version of http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 with the import having a ^^xs:string 10:42:52 <trackbot> Sorry, amibiguous username (more than one match) - jos 10:42:52 <trackbot> Try using a different identifier, such as family name or username (eg. jdebruij2, jderoo) 10:43:01 <sandro> ACTION: jdb to add a version of http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 with the import having a ^^xs:string 10:43:02 <trackbot> Sorry, couldn't find user - jdb 10:43:15 <sandro> ACTION: jdebruij2 to add a version of http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 with the import having a ^^xs:string 10:43:15 <trackbot> Created ACTION-684 - to add a version of http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_2 with the import having a ^^xs:string [on Jos de Bruijn - due 2009-01-21]. 10:44:12 <sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_3 with the same caveats as http://www.w3.org/2005/rules/wiki/RDF_Combination_Constant_Equivalence_Graph_Entailment_2 10:44:17 <DaveReynolds> -1 10:44:32 <StellaMitchell> I changed the language label 10:45:16 <sandro> Yeah, you shouldn't use rdf:text in Turtle. 10:47:41 <josb> So, you would propose not to define combination with RDF graphs that use rdf:text? 10:48:30 <sandro> Yes 10:49:26 <sandro> dave: it's fine for rdf:text to be in the RIF part. 10:50:39 <Harold> Dave: Must that you do the mapping back again into RDF. 10:51:34 <josb> Again, so, you would propose not to define combination with RDF graphs that use rdf:text? 10:51:39 <sandro> csma: this ( ex:a ex:p "with language tag@en"^^rdf:text . ) is bad syntax, as I'm understanding it. 10:55:22 <Harold> Jos: propose to extend issue 90 to rdf:text. <sandro> SubTopic: RDF_Combination_Member_1 10:56:07 <sandro> http://www.w3.org/2005/rules/wiki/RDF_Combination_Member_1 11:01:01 <sandro> PROPOSED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Member_1 for all dialects (BLD, Core, CoreSafe, PRD) 11:01:16 <DaveReynolds> +1 11:01:17 <GaryHallmark> +1 11:01:18 <sandro> +1 11:01:20 <Harold> +1 11:01:20 <AdrianP> +1 11:01:22 <Hassan> +1 11:01:23 <Michael_Kifer> +1 11:01:25 <josb> +1 11:01:31 <sandro> RESOLVED: Approve http://www.w3.org/2005/rules/wiki/RDF_Combination_Member_1 for all dialects (BLD, Core, CoreSafe, PRD) 11:01:33 <StellaMitchell> Three are 3 more proposed RDF combination test cases, and you can find them here: http://www.w3.org/2005/rules/wiki/Category:RDF%2BBLD_Test 11:02:11 <Harold> Sandro: Loved to see another test case showing ## vs. subclass. 11:02:33 <sandro> ACTION: jos to write a test case or two to illiustrate how ## and rdfs:subClassOf relate 11:02:33 <trackbot> Sorry, amibiguous username (more than one match) - jos 11:02:33 <trackbot> Try using a different identifier, such as family name or username (eg. jdebruij2, jderoo) 11:02:44 <sandro> ACTION: jdebruij2 to write a test case or two to illiustrate how ## and rdfs:subClassOf relate 11:02:44 <trackbot> Created ACTION-685 - Write a test case or two to illiustrate how ## and rdfs:subClassOf relate [on Jos de Bruijn - due 2009-01-21]. 11:03:11 <Zakim> -Hassan_Ait-Kaci 11:03:11 <sandro> csma: Obviously, these '##' test cases don't apply to Core if ## isn't in Core. <sandro> Topic: Morning Break 11:03:36 <Zakim> -DaveReynolds 11:03:37 <Zakim> -Stella 11:03:48 <Zakim> -LeoraMorgenstern 11:04:49 <sandro> action: josb remind Sandro that 'josb' works in assigning actions. 11:04:50 <trackbot> Created ACTION-686 - Remind Sandro that 'josb' works in assigning actions. [on Jos de Bruijn - due 2009-01-21]. 11:05:06 <sandro> action: jdb remind Sandro that 'jdb' works in assigning actions. 11:05:06 <trackbot> Sorry, couldn't find user - jdb 11:05:17 <sandro> trackbot, reload 11:05:48 <sandro> action: jdb remind Sandro that 'jdb' ALSO works in assigning actions. 11:05:49 <trackbot> Sorry, couldn't find user - jdb 11:29:56 <Zakim> +DaveReynolds <sandro> Topic: Object representation (and relationship with frames) 11:31:13 <Zakim> +LeoraMorgenstern 11:42:05 <pvincent> scribenick pvincent 11:42:14 <Zakim> +[IBM] 11:42:40 <StellaMitchell> zakim, ibm is temporarily me 11:42:40 <Zakim> +StellaMitchell; got it 11:42:56 <pvincent> F2F12 late am agenda item = 11:30-12:30 - Object representation (and relationship with frames) 11:45:04 <pvincent> CSMA: presentation on Objects vs Frames - how to translate to/from an OO rule language and RIF frames 11:45:35 <pvincent> CSMA: problem is translating back and forth, where there is no external object model / schema 11:45:37 <sandro> csma: not a problem when you have a schema.... 11:46:45 <sandro> slide- Assert(Obj[att->val2]) ----> Obj.att=(val1 val2) 11:48:24 <sandro> Harold: We had a proposal for lists that was very close, but didn't quite make it into BLD. 11:49:22 <pvincent> CSMA: only remaining problem: no way to specify if a frame value is single value or multi value 11:50:45 <sandro> Two problems: losing MaxCardinality, and no-support-for-lists 11:52:19 <pvincent> CSMA: one solution is cardinality etc metadata but this implies inter-dialect problems 11:53:53 <sandro> Michael_Kifer: This would mean we'd have to have equality in the head... 12:04:56 <pvincent> The problem being discussed: factA and factB existence in Core / logic may be replaced in PRD with factA is replaced with factB (A = B)... 12:06:08 <sandro> Michael_Kifer: We could use metadata to convey cardinality. 12:10:52 <pvincent> CSMA: adds Obj[(comment: Card=1) att->val] 12:11:19 <sandro> Gary: Two different problems: (1) How do you convey the maxCard, and (2) what do you do with the infomaiton? 12:11:49 <AdrianP> in BLD o[a->b a->b] reduces to o[a->b a->b] 12:12:06 <DaveReynolds> If the answer to (2) is "it changes how you transform to/from your object model but not the semantics of the rules" then the answer to (1) could indeed be "use metadata" 12:12:10 <AdrianP> but it could be a list collection o[a->list(b,b)] 12:13:23 <AdrianP> in BLD it reduces to o[a->b] 12:20:20 <sandro> pvincent, your scribing is..... behind. :-) 12:24:34 <Harold> Put it into the Conformance Clauses? 12:26:20 <Harold> E.g., a slot-uniqueness-conformant BLD?,? consumer/producer ... 12:27:24 <DaveReynolds> Any chance of transcribing something of what you are talking about so we can see it? 12:30:04 <GaryHallmark> test case: car[color->red color->blue] is ok but car[color=red color=blue] is not ok 12:31:23 <pvincent> MK: SANDRO_CAR[metadata Color -> Red] vs SANDRO_CAR[Color=red] 12:31:29 <DaveReynolds> So does Group( car[color=red] Forall ?x (?x[color=blue]) :- (?x(color=red)) ) entail car[color=red] ? 12:33:27 <pvincent> CSMA: suggestion is add cardinality metadata in PRD; then if you want Core compliance you can't use Obj[att=val] only Obj[att->val] 12:40:38 <sandro> csma: just use some special syntax (eg =) in PRD. 12:43:41 <sandro> csma: So PRD implementations should use "=" when they need to, and "->" when they don't need to, and want interop with Core/BLD. 12:43:44 <pvincent> CSMA: suggesting... PRD supports "=" which does not translate to core, as well as "->"; "=" is not translatable to core as it implies an overwrite 12:44:53 <sandro> Sandro: "=' doesn't just mean maxCard=1, it ALSO implies over-writing. 12:46:26 <sandro> csma: No, we'll use UPDATE action for that 12:46:41 <sandro> Sandro: But, but, but, you just told me.... 12:48:22 <DaveReynolds> If PRD does not need the maxCard case to imply update semantics, it certainly doesn't imply equality, so in that case it *is* just annotations to keep track of how to map back results to objects in which case it could be in Core. 12:49:20 <sandro> xPROPOSED- PRD extends frames with some syntactic indicator of maxCard=1 and that over-writing should be done when additional values are encountered, to be used only in rule conclusions. ....... 12:50:09 <sandro> the semantics only matter in the conclusion, but for translations, you need to know about it when you do conditions, maybe? 12:52:55 <sandro> PROPOSED: PRD extends frames with some new syntax for attributes which are single-valued and have replacement semantics (matching OO object attributes). This doesn't make sense for Core or BLD. 12:53:32 <sandro> Michael_Kifer: Maybe prolog could be changed to do this, maybe.... 12:53:44 <sandro> PROPOSED: PRD extends frames with some new syntax for attributes which are single-valued and have replacement semantics (matching OO object attributes). This doesn't make sense for Core or BLD. 12:53:59 <AdrianP> in Prolog you have lists which can be used to have multiple values 12:54:53 <sandro> (proposal tabled for now.... no objections, but maybe a little more clarification, esp w.r.t. schemas.) <sandro> summary: Consensus was almost (but not) reached, that PRD should be extended with a new syntax for indicating that certain slots are single-valued and have "replacement semantics". (Topic revisited on Day 2.) <sandro> topic: Lunch Break 12:55:19 <sandro> ADJOURN until 12:45 12:55:34 <Zakim> -DaveReynolds 12:55:39 <Zakim> -StellaMitchell 12:55:41 <Zakim> -LeoraMorgenstern 12:56:41 <sandro> ADJOURN until 1:45 13:01:50 <sandro> Harold, e-mail to www-archive@w3.org and then look at http://lists.w3.org/Archives/Team/w3t-archive/2009Jan/ for it to show up. Then link to it. 13:26:22 <sandro> Harold, here: http://lists.w3.org/Archives/Public/www-archive/2009Jan/0005.html 13:43:54 <Zakim> +DaveReynolds 13:58:36 <sandro> zakim, who is on the phone? 13:58:37 <Zakim> On the phone I see Mike_Dean, RIF_Meeting_Room, DaveReynolds 13:58:38 <Zakim> RIF_Meeting_Room has Christian, Gary, Harold, Adrian, Jos, Paul, Micheal_Kifer, Changhai, Sandro <sandro> Topic: Interoperability with XML (issue-37 and issue-38) 13:58:42 <sandro> starting again. 14:00:50 <sandro> http://www.w3.org/2005/rules/wg/track/issues/37 14:00:56 <sandro> http://www.w3.org/2005/rules/wg/track/issues/38 14:01:01 <PaulVincent> scribenick PaulVincent 14:01:11 <sandro> scribenick: PaulVincent 14:01:35 <PaulVincent> First afternoon topic: Interoperability with XML 14:02:19 <sandro> http://lists.w3.org/Archives/Public/public-rif-wg/2008Oct/0046.html 14:04:46 <sandro> group is looking at http://lists.w3.org/Archives/Public/public-rif-wg/2008Oct/0046.html 14:05:06 <PaulVincent> Gary: look at example (see link above) 14:05:22 <DaveReynolds> I 14:05:38 <DaveReynolds> Gary - could you speak closer to a mike? 14:06:42 <sandro> better? 14:06:52 <DaveReynolds> Slightly, thanks. 14:07:30 <PaulVincent> Sandro: syntax on curies eg <tns:/shiporder> should not have angled brackets around them as they are not IRIs 14:08:28 <PaulVincent> Gary: This has as simple standard frame naming from schema definitions 14:11:30 <PaulVincent> Note- this XML doc to frame mapping is for the convenience of understanding RIF syntax versus XML docs... it would not be necessary to consider this at runtime (when interchanging rules against XML docs...) 14:16:23 <PaulVincent> Jos: can the full identifier such as /shiporder/item/title not just be replaced by title? 14:17:03 <PaulVincent> Gary: problem is these named elements can be repeated with different types: element identifier is their path 14:17:30 <PaulVincent> Gary: missing info in this scheme includes order, cardinality, ... 14:29:57 <DaveReynolds> This proposal would also be compatible with using sawsdl to annotate the schema which a URI for the types and properties if you want to. 14:30:09 <DaveReynolds> s/which/with/ 14:31:27 <PaulVincent> Gary: Equivalent to JAXB mapping 14:38:47 <sandro> Jos: I still don't understand why you want the whole path, but we can move on. 14:41:55 <PaulVincent> Discussion of Gary's General Rules for XML-Frame mapping ... 14:42:17 <PaulVincent> CSMA: points 1-3 are still available from the schema. 14:43:00 <DaveReynolds> Jos - the names in XML schema are context sensitive, somewhat like nested structs in C or nested static classes in Java. So property foo on complex type A can, for example, have a different range from property foo on complex type B. They are not the "same" foo so when assigning URIs making them different URIs makes sense. 14:43:09 <Zakim> +ChrisW 14:44:03 <PaulVincent> CSMA: ... hence that a RIF engine has different requirements for XML-to-frame versus an interchange for engines that will use the schema natively... 14:47:02 <DaveReynolds> Are you talking about xml schema component designators? 14:47:15 <DaveReynolds> http://www.w3.org/TR/xmlschema-ref/ 14:47:18 <sandro> XML Schema Component Designators. 14:49:26 <Zakim> -ChrisW 14:50:01 <Zakim> +ChrisW 14:51:03 <sandro> Adrian: What about xs:any 14:51:17 <sandro> Sandro: ... then you're in the land of schema-free XML, and none of this applies. 14:51:35 <PaulVincent> Discussion on issues with Garys proposal 14:52:40 <PaulVincent> CSMA: some of these are for translation of XML docs into RIF which is not required for all RIF use cases (eg where PRD is running rules in an external engine) 14:53:27 <PaulVincent> Sandro: # and ## for ground facts should be in core 14:54:14 <DaveReynolds> Sandro - you may have to fight for that. The divergence between ## and rdfs causes me problems I'm trying to keep ## out of Core :-) 14:55:30 <PaulVincent> Move to CSMA proposal at http://www.w3.org/2005/rules/wiki/RIF%2BXML_data-schema 14:56:38 <DaveReynolds> For the record: on Gary's proposal, the rules don't explicitly say that the simple type mapping is done but it should be. So in the example the price value would map to "9.90"^^xsd:decimal. 14:56:52 <sandro> +1 DaveReynolds 14:59:37 <PaulVincent> CSMA discussing Named element selection: named descendent 15:00:30 <sandro> csma: note that I used xs:double cast in the first example, since it's schema-less. 15:02:01 <Zakim> +Hassan_Ait-Kaci 15:02:55 <PaulVincent> CSMA discussing sub-element query: child 15:05:16 <PaulVincent> CSMA now discussing attribute query: attribute 15:07:32 <DaveReynolds> I'm going to sign off in a moment. 15:07:50 <PaulVincent> Goodnight, Dave 15:08:22 <PaulVincent> CSMA discusses With an XML Schema for the target XML document 15:08:41 <sandro> any thoughts on this stuff, DaveReynolds ? 15:08:42 <DaveReynolds> One comment on the approach which I'll leave with you :-) Does this proposal assume well -striped XML? The use of # suggests that it does. 15:09:00 <Zakim> -DaveReynolds 15:13:53 <PaulVincent> CSMA discusses Schema-element selection: schema-element descendant 15:20:15 <sandro> Jos: Can csma's proposal be done just by providing a predicate for xpath? 15:22:07 <sandro> Jos: It would avoid this odd mixing of syntaxes 15:23:11 <sandro> csma: eh then the whole query woul dhave to be in xpath 15:23:50 <Zakim> -ChrisW 15:24:30 <Zakim> +ChrisW 15:26:15 <Zakim> -ChrisW 15:27:26 <Zakim> +ChrisW 15:36:05 <sandro> csma: on board ---- ?x # ex:/descendant::shiporder 15:36:24 <PaulVincent> CSMA: X # FX:/DESCENDENT::SHIPORDER 15:38:22 <Zakim> -ChrisW 15:39:10 <sandro> csma: My proposal --- ?x @ tns:/descendant::shiporder ?x[child::shipto->?y] 15:40:13 <sandro> csma: Gary's Proposal --- ?x # tns:shiporder DEEP: ?x[tns:/shiporder/shipto -> ?y ] FLAT: ?x[tns:/shipto->?y] 15:42:03 <Zakim> -Mike_Dean 15:42:51 <sandro> Sandro: test case --- you cannot quanltify over slot names in csma's proposal. 15:43:43 <sandro> sandro: csma, you're providing a bunch of external/magic slot names and types. 15:44:14 <sandro> jos: Gary's doesn't need/use externals. csma's uses externals. 15:46:40 <sandro> csma: ... and ?y[child::name->"Ola Nordmann"] 15:47:43 <sandro> csma: equivalent to ?x[child::shipto/child::name->""Ola Nordmann"] and we could start from root. 15:50:51 <sandro> jos: how about a general embedding? every XML document mapped to a RIF formula? 15:51:08 <sandro> sandro: Yeah, you could use a trimmed down version of the PSVI (post-schema-validation infoset). 15:55:11 <sandro> sandro: Is something like this necessary for PRD's success? 15:55:59 <sandro> changhai: I think so; I think I like the flexibility in the direction of csma's proposal.... 15:56:52 <sandro> ... I'm concerned about how we get collections of values into RIF. 15:58:05 <sandro> csma: Use of type information from xml schema is included explicitely in mine. I think you can use it the same way in Gary's. 16:00:13 <sandro> Changhai: in foo # X X is a type, not a collection.... 16:04:18 <sandro> Sandro: Is this work necessary to PRD? 16:04:46 <sandro> Paul: Yes, althoug maybe we need even better access to XML (than frames allow). 16:06:05 <sandro> Gary:My proposal is pretty much how our product works (jaxb style -- flash conversion). The XPath peek & poke approach would be something we'd have to hide in translation. 16:07:05 <sandro> Gary: I'm not wedded to using the schema to construct names. JAXB just uses the local names. 16:07:51 <sandro> sandro: If RIF doesn't do this, you'll need to do a non-standard version of this? 16:08:00 <sandro> csma, paul, gary: yes. 16:08:41 <sandro> paul: PRR dodges this issue. 16:08:59 <sandro> Adrian: reaction-ruleml has a way to do something like this. 16:10:02 <sandro> csma: You could re-use this work with UML instead of Schema. 16:11:08 <sandro> "Using RIF with XML and Object-Oriented Data" ? 16:11:24 <sandro> csma: I think we'll be content with just "Using RIF with XML" 16:13:45 <AdrianP> additionally in a query /constraint expression language you could explicitly define closures over object relations which can not be done directly in FOL 16:14:48 <sandro> csma: If you want to *really* retrieve a class and it's subclasses, you need to do it several different ways, given XMLS. 16:15:36 <PaulVincent> Way ahead suggestion: have 1 or more DSLs evaluate these approaches for embedding XML-enabled PRD in their docs, eg MISMO? 16:19:48 <sandro> PROPOSED: Create a new document, something like "Using RIF with OWL", addressing ISSUE-37 and ISSUE-38, in the space of Gary and Christian's proposals, for RIF rules to use XML data wth and without a schema. What it specifies should work with Core, BLD, and PRD, and is important to the success of RIF. 16:20:05 <sandro> gah. typo of "OWL" for "XML" :-) 16:20:29 <sandro> PROPOSED: Create a new document, something like "Using RIF with XML", addressing ISSUE-37 and ISSUE-38, in the space of Gary and Christian's proposals, for RIF rules to use XML data wth and without a schema. What it specifies should work with Core, BLD, and PRD, and is important to the success of RIF. 16:20:33 <Harold> zakim, who is here? 16:20:33 <Zakim> On the phone I see RIF_Meeting_Room, Hassan_Ait-Kaci (muted) 16:20:34 <Zakim> RIF_Meeting_Room has Christian, Gary, Harold, Adrian, Jos, Paul, Micheal_Kifer, Changhai, Sandro 16:20:36 <Zakim> On IRC I see StellaMitchell, ChrisW, Hassan, csma, PaulVincent, josb, GaryHallmark, Michael_Kifer, cke, Harold, RRSAgent, AdrianP, sandro, mdean, trackbot, Zakim 16:21:43 <sandro> csma: it's important that whatever we do relies on features in current engines. and be extensible for doing stuff like more of XPath in the future. 16:21:45 <Zakim> +Stella_Mitchell 16:22:04 <sandro> PROPOSED: Create a new document, something like "Using RIF with XML", addressing ISSUE-37 and ISSUE-38, in the space of Gary and Christian's proposals, for RIF rules to use XML data wth and without a schema. What it specifies should work with Core, BLD, and PRD, and is important to the success of RIF. 16:22:50 <ChrisW> i haven't been part of the discussion 16:23:03 <ChrisW> but i would strongly object to starting new work 16:23:15 <sandro> PROPOSED: Create a new document, something like "Using RIF with XML", addressing ISSUE-37 and ISSUE-38, in the space of Gary and Christian's proposals, for RIF rules to use XML data wth and without a schema. What it specifies should work with Core, BLD, and PRD, and is important to the success of RIF. This technology is not just for use of XML data, but for using XML Schemas to communicate data models. 16:23:54 <sandro> ChrisW, Gary, Paul, and the ILOG (and IBM company) folks say they have to have this for PRD to be useful. 16:24:04 <sandro> s/,/-/ 16:24:49 <ChrisW> I've heard the same argument for a lot of things in RIF 16:25:39 <csma> The alternative to starting a new doc is including it in PRD. 16:26:13 <csma> since this is useful to other dialects as well, better put it in a separate doc instead of a PRD section 16:26:49 <ChrisW> finish the work on PRD first, then consider this for a note. But only when the existing drafts are at LC 16:27:05 <ChrisW> no new work 16:27:09 <ChrisW> i repeat, no new work 16:27:10 <sandro> sandro: I think the risk of new-work is somewhat reduced buy having it be an independent document. 16:27:31 <Harold> Also look into related work such as Schema-Aware Queries and Stylesheets (http://www.stylusstudio.com/schema_aware.html) and Schema-aware processing with XSLT (http://www.ibm.com/developerworks/xml/library/x-schemaxslt.html). 16:27:47 <ChrisW> this has too many other implications 16:27:55 <ChrisW> i will object 16:30:44 <sandro> sandro: We could publish a VERY DRAFT WD on this, which is just the cut and paste of Gary's and CSMA's proposals, to [as csma says] let people know this kind of work is doable, so they know PRD might be useful. 16:31:31 <sandro> gary: In our product, we tried an xpath thing, but ended up putting the whole document in, because it worked better for what customers wanted. 16:32:58 <sandro> csma: I want to borrow XPath. 16:34:33 <sandro> gary: RIF CL is already a query language, so it's kind of painful to combine it with another, like XPath. 16:34:35 <sandro> csma: true. <sandro> topic: Afternoon Break 16:34:42 <sandro> BREAK until 17:00 16:35:06 <Zakim> -Stella_Mitchell 16:37:21 <Zakim> -Hassan_Ait-Kaci 17:04:53 <sandro> scribenick: Michael_Kifer <sandro> topic: Forall in Core/BLD vs PRD 17:05:02 <sandro> restarting 17:06:31 <Michael_Kifer> CSMA: issue: forall is different in BLD and PRD 17:06:47 <sandro> csma: Forall means something different in PRD from Core. It means "for each value that you can bind" ... which has the same effect. But PRD also needs the real forall...? 17:10:18 <sandro> sandro: (listening to Changhai's example) You want an aggregate. if forall x... 17:10:33 <Zakim> +Stella_Mitchell 17:13:31 <sandro> sandro: so this, in practice, needs closed world assumption, 17:13:35 <sandro> jos: or negation 17:14:37 <sandro> gary: we have this now in PRD via "not exists". 17:15:56 <Michael_Kifer> the issue seems to be "forall" outside of the rule vs. "forall" in the rule body 17:16:17 <sandro> paul: "if at least 3..." 17:17:01 <sandro> csma: So the forall inside the rule and outside the rule are really the same? 17:17:10 <sandro> Michael_Kifer: Yes 17:17:15 <sandro> Jos: Yes 17:21:55 <Michael_Kifer> decided that this is a non-issue 17:21:56 <Harold> BLD's Forall can work over infinite sets such as {0, s(0), s(s(0)), . . .}: 17:22:01 <Harold> Forall ?N ( nat(s(?N)) :- nat(?N) ) 17:22:30 <josb> Next topic, please! 17:22:52 <sandro> csma: on board: 17:22:52 <sandro> forall ?x ( ?x # Car) 17:22:52 <sandro> if forall ?y ( ?y # Door ) 17:22:52 <sandro> ?x[door->?y] and ?y[color->red] 17:22:52 <sandro> then Paint It Blue 17:23:13 <sandro> (all agree, the two forall as "the same", and can use the same syntax.) 17:23:42 <sandro> (if the language were to support forall inside a condition.) 17:23:45 <Michael_Kifer> Topic: issue 39, RIF should support import or inclusion of rulesets 17:23:46 <josb> http://www.w3.org/2005/rules/wg/track/issues/39 17:27:00 <Michael_Kifer> import is in core already. CSMA: might cause conflicts in conflict resolution strategies 17:28:56 <sandro> agreed. If PRD imports with a different Res.Strat. it should ... say what happens. (one rules, or error, or something.) 17:29:10 <Michael_Kifer> proposed: close issue 39 17:29:24 <sandro> PROPOSED: Close issue-39, saying ruleset-imports is in Core, and we're not going to define an "includes" at this time. 17:29:35 <Michael_Kifer> because core already has import 17:29:36 <josb> +1 17:30:02 <Harold> +1 17:30:34 <sandro> Michael_Kifer: i'd like includes would be including rules, not whole documents.... 17:30:55 <sandro> PROPOSED: Close issue-39, saying ruleset-imports is in Core, and we're not going to define an "includes" at this time (in part because we don't know what it might mean). 17:31:02 <sandro> +1 17:31:07 <josb> ++1 17:31:17 <Michael_Kifer> +1 17:31:18 <josb> +++1 17:36:32 <sandro> Sandro: it sounds like Core:imports wont be a pain to implement in PRD, even though it doesn't do all the fance stuff some PRD engines really want in imports. 17:37:50 <sandro> csma: Import is specified in the dumb way -- without juggling priorities -- so it's not very useful in PRD. 17:38:35 <Zakim> +Hassan_Ait-Kaci 17:39:01 <sandro> gary: Yeah, someday we'll need to specify complex ways to do import in PRD. 17:39:53 <Michael_Kifer> current import mechanism is too primitive for PRD. Need to expand it in the future for PRD 17:41:01 <sandro> RESOLVED: Close issue-39, saying ruleset-imports is in Core, and we're not going to define an "includes" at this time (in part because we don't know what it might mean). It ends up in PRD, where it's not ideal, but not really harmful. Some PRD may do a more sophisticated import at some point. 17:41:03 <Michael_Kifer> csma: the current import mechanism in core is acceptable, but it is not very interesting for prd. 17:41:28 <sandro> s/RESOLVED/PROPOSED/ 17:42:15 <sandro> PROPOSED: Close issue-39, saying ruleset-imports is in Core, and we're not going to define an "includes" at this time (in part because we don't know what it might mean). 'Imports' ends up in PRD, where it's not ideal, but not really harmful. Some version of PRD may do a more sophisticated import at some point. 17:42:26 <sandro> +1 17:42:47 <sandro> Harold: when you find the better imports, propose it to the rest of RIF! 17:42:58 <Harold> +1 17:43:05 <josb> +1 17:43:06 <Michael_Kifer> +1 17:43:06 <AdrianP> +1 17:43:07 <GaryHallmark> +1 17:43:16 <PaulVincent> +1 17:43:45 <sandro> RESOLVED: Close issue-39, saying ruleset-imports is in Core, and we're not going to define an "includes" at this time (in part because we don't know what it might mean). 'Imports' ends up in PRD, where it's not ideal, but not really harmful. Some version of PRD may do a more sophisticated import at some point. 17:43:55 <sandro> RRSAgent, pointer? 17:43:55 <RRSAgent> See http://www.w3.org/2009/01/15-rif-irc#T01-44-04 17:50:35 <sandro> Sandro: let's put "filters" ("such that" expressions) as syntactic sugar into Core. 17:50:52 <sandro> MK: bounded quantifies. 17:50:53 <cke> good idea 17:50:58 <josb> Let's not. The syntax is already complicated enough 17:52:20 <sandro> Sandro: So, if we have to do a LC2 -- then we can talk about this. 17:54:09 <sandro> ISSUE: Should we put bounded quantifiers into Core? (forall x such than x # C ...) (esp if we need an LC2 for BLD anyway) 17:54:23 <trackbot> Created ISSUE-91 - Should we put bounded quantifiers into Core? (forall x such than x # C ...) (esp if we need an LC2 for BLD anyway) ; please complete additional details at http://www.w3.org/2005/rules/wg/track/issues/91/edit . 17:57:52 <sandro> MK: bounded quants are not very useful in BLD, and the re-writing is a little tricky. 17:58:42 <sandro> csma: Many rules are written with BQs, and people don't want to discard that optimization. Also, you need BQs for any else-clause. 17:59:18 <Zakim> -Stella_Mitchell 17:59:31 <Michael_Kifer> Forall ?X in C (h :- body) is Forall ?X (h :- body and C). 18:01:40 <sandro> ADJOURN - 1 minute late 18:02:33 <josb> http://maps.google.com/maps?f=q&source=s_q&hl=en&geocode=&q=higgins+portland&sll=37.0625,-95.677068&sspn=34.945679,58.886719&ie=UTF8&ll=45.51541,-122.680957&spn=0.001887,0.003594&z=18&iwloc=A 18:04:01 <josb> http://maps.google.com/maps?f=d&source=s_d&saddr=SW+5th+Ave&daddr=1239+SW+Broadway,+Portland,+OR+97205+(Higgins+Restaurant)&hl=en&geocode=FYaBtgId7w6w-A%3B&mra=cc&dirflg=w&sll=45.515407,-122.680786&sspn=0.001887,0.003594&ie=UTF8&z=18 18:06:39 <josb> coffee is at http://maps.google.com/maps?f=d&source=s_d&saddr=45.515221,-122.679648&daddr=128+SW+3rd+Ave,+Portland,+OR+97204+(Stumptown+Coffee+Roasters)&hl=en&geocode=&mra=mi&mrsp=0&sz=15&dirflg=w&sll=45.518045,-122.676859&sspn=0.015095,0.028753&ie=UTF8&z=15 18:08:22 <Zakim> -Hassan_Ait-Kaci 18:09:21 <Zakim> -RIF_Meeting_Room 18:09:22 <sandro> zakim, who is on the phone? 18:09:23 <Zakim> SW_RIF(F2F12)11:00AM has ended 18:09:25 <Zakim> Attendees were Mike_Dean, Hassan_Ait-Kaci, Christian, Gary, Harold, Adrian, Jos, Paul, Micheal_Kifer, Changhai, Sandro, Stella, +44.145.441.aaaa, DaveReynolds, +1.212.781.aabb, 18:09:29 <Zakim> ... LeoraMorgenstern, StellaMitchell, ChrisW, Stella_Mitchell 18:09:31 <Zakim> apparently SW_RIF(F2F12)11:00AM has ended, sandro 18:09:33 <Zakim> On IRC I see cke, ChrisW, Hassan, csma, josb, GaryHallmark, Michael_Kifer, Harold, RRSAgent, sandro, mdean, trackbot, Zakim 18:14:56 <sandro> rrsagent, make record public 18:14:57 <RRSAgent> I have made the request, sandro 18:15:01 <sandro> rrsagent, make minutes 18:15:01 <RRSAgent> I have made the request to generate http://www.w3.org/2009/01/15-rif-minutes.html sandro