IRC log of sw-team on 2003-04-29
Timestamps are in UTC.
- 17:59:52 [RRSAgent]
- RRSAgent has joined #sw-team
- 17:59:58 [Zakim]
- SW_SWAD()2:00PM has now started
- 17:59:59 [Zakim]
- +DanC
- 18:00:17 [Zakim]
- +DanBri
- 18:00:33 [JosD]
- JosD has joined #sw-team
- 18:00:37 [Zakim]
- -DanBri
- 18:00:46 [reagleMIT]
- reagleMIT has joined #sw-team
- 18:01:03 [Zakim]
- +DanBri
- 18:01:38 [Zakim]
- +Marja
- 18:01:44 [Zakim]
- +??P8
- 18:01:59 [Zakim]
- +??P9
- 18:02:07 [Zakim]
- +TimBL
- 18:02:30 [ryan-lap]
- ryan-lap has joined #sw-team
- 18:02:38 [sandro]
- zakim, who is here?
- 18:02:38 [Zakim]
- On the phone I see DanC, DanBri, Marja, JosD, ??P9, TimBL
- 18:02:39 [Zakim]
- On IRC I see ryan-lap, reagleMIT, JosD, RRSAgent, Zakim, marja, danbri, RalphS, Nobu, emscribe, dom, DanC, sandro, ericP, Ralph
- 18:02:47 [Danny-DC]
- Danny-DC has joined #sw-team
- 18:03:18 [sandro]
- Danny, are you joining us today?
- 18:03:19 [tim-mit]
- tim-mit has joined #sw-team
- 18:03:27 [Danny-DC]
- yes, I am.
- 18:04:20 [DeborahMc]
- DeborahMc has joined #sw-team
- 18:04:53 [JosD]
- JosD has joined #sw-team
- 18:04:53 [tim-mit]
- tim-mit has joined #sw-team
- 18:04:53 [DeborahMc]
- DeborahMc has joined #sw-team
- 18:04:58 [danbri]
- odd...
- 18:05:01 [DanC]
- Zakim, Stanford/KSL holds DebM, Paulo
- 18:05:01 [Zakim]
- +DebM, Paulo; got it
- 18:05:15 [DanC]
- Zakim, who's on the phone?
- 18:05:15 [Zakim]
- On the phone I see DanC, DanBri, Marja, JosD, Stanford/KSL, TimBL
- 18:05:16 [Zakim]
- Stanford/KSL has DebM, Paulo
- 18:05:28 [Zakim]
- +Danny
- 18:05:49 [tim-mit]
- Zakim, TimBL is really TimsOffice
- 18:05:49 [Zakim]
- +TimsOffice; got it
- 18:06:09 [tim-mit]
- Zakim, TimsOffice holds TimBL, Sandro, Ryan, EricP
- 18:06:09 [Zakim]
- +TimBL, Sandro, Ryan, EricP; got it
- 18:07:02 [Zakim]
- +EMiller
- 18:07:55 [sandro]
- RRSAgent, pointer?
- 18:07:55 [RRSAgent]
- See http://www.w3.org/2003/04/29-sw-team-irc#T18-07-55
- 18:08:40 [DanC]
- [[ The IRC log will
- 18:08:40 [DanC]
- appear at http://www.w3.org/2003/04/29-sw-team-irc. ]]
- 18:09:16 [DanC]
- ------- convene
- 18:09:19 [DanC]
- proposed agenda: http://www.w3.org/2003/04/29/swad/
- 18:09:58 [ericP-scribe]
- sandro: what do we do with the irc log of this conference?
- 18:10:11 [ericP-scribe]
- debM: public is OK with us
- 18:10:18 [ericP-scribe]
- [no objections]
- 18:10:22 [sandro]
- RESOLVED record of this meeting be public
- 18:10:50 [ericP-scribe]
- q+ to be an example
- 18:11:07 [sandro]
- ack ericP-scribe
- 18:11:07 [Zakim]
- ericP-scribe, you wanted to be an example
- 18:12:18 [ericP-scribe]
- sandro: this is DAML-funded research, _not_ rec track or member-supported
- 18:12:40 [ericP-scribe]
- sandro: we'll discuss follow up meetings in the last 15 mins of this meeting
- 18:12:51 [danbri]
- (although we hope some Members are supportive of these efforts... they're just not funding it via Member fees)
- 18:13:11 [ericP-scribe]
- [tim returns]
- 18:13:40 [ericP-scribe]
- sandro: goal: to plan and design for interop between planners, reasoners and explainers
- 18:14:52 [DeborahMc]
- debm cant hear either
- 18:15:10 [ericP-scribe]
- debM, can you hear sandro?
- 18:15:20 [DeborahMc]
- yes sorry - it was just eric
- 18:15:33 [ericP-scribe]
- danc: proving to humans is interesting, but a side goal
- 18:15:48 [DanC]
- *not* a goal, to me, to prove things to human
- 18:15:54 [DanC]
- might be a side benefit
- 18:15:58 [ericP-scribe]
- sandro: debM's interest is in explainer UIs, nto verifiers
- 18:16:37 [ericP-scribe]
- sandro: if PPF works for a verifier, that's just gravy?
- 18:17:05 [ericP-scribe]
- debM: yes, one datapoint of buy-in from the verifyer community
- 18:17:11 [sandro]
- http://www.w3.org/2003/04/29/swad/
- 18:18:13 [ericP-scribe]
- debM: may want to add use cases, perhaps in follow-up email
- 18:18:43 [ericP-scribe]
- sandro: coordinating multiple reasoners, JTP and cwm work on common proof
- 18:19:06 [tim-mit]
- exactly
- 18:19:13 [ericP-scribe]
- paulo: if reasoner B uses reasoner A, it can use the PPF context
- 18:19:45 [ericP-scribe]
- jos: +testing 2nd opinions by idependent reasoners
- 18:19:55 [DanC]
- 2nd opinion test cases... pointer? could you follow up in email with pointers?
- 18:20:56 [ericP-scribe]
- paulo: multiple justifications makes proof more complex but allows one to detect contradictions
- 18:21:18 [DanC]
- agenda + reductio ad absurdum, LEM (under C. Issues)
- 18:21:34 [JosD]
- will do that (but proofs are kind of not very human readable...)
- 18:21:43 [DanC]
- agenda + how registration works with registered stuff (under C. Issues)
- 18:21:54 [ericP-scribe]
- paulo: general plan it to focus on justification
- 18:22:09 [ericP-scribe]
- debm: today we provide a proof browser.
- 18:22:35 [ericP-scribe]
- ... using rewrite rules to transform proof trace into an explaination, abstracting away some details.
- 18:22:56 [ericP-scribe]
- ... could make it more understandable for any particular audience.
- 18:23:35 [ericP-scribe]
- paulo: one way of seeing how a machine arrived at a proof is by verifying the proof
- 18:23:39 [JosD]
- there is the pointer to the original as well
- 18:23:44 [DanC]
- proposed proof language in DAML: http://www.ksl.stanford.edu/software/IW/spec/iw.daml not sure if there's a prose write-up. but that's what I found from the agenda.
- 18:23:50 [sandro]
- http://belo.stanford.edu:8080/iwbrowser/WffBrowser?url=http://www.ksl.stanford.edu/software/IW/sample/TS/IW1.daml
- 18:24:28 [ericP-scribe]
- sandro: i found it easiest to play with browser making chagnes and examining RDF.
- 18:24:33 [ericP-scribe]
- ... do you have a prepared path?
- 18:24:45 [ericP-scribe]
- debm: planned to be responsive to your needs.
- 18:24:56 [ericP-scribe]
- [debm walks us trhough tony is a seafood]
- 18:26:03 [ericP-scribe]
- danc, it's all onclick events
- 18:26:36 [sandro]
- http://belo.stanford.edu:8080/iwbrowser/WffBrowser?url=http://www.ksl.stanford.edu/software/IW/sample/TS/IW1.daml
- 18:27:59 [ericP-scribe]
- [viewing with proof style: proof, sentence format: english, lens mag: 3
- 18:28:00 [ericP-scribe]
- ]
- 18:29:02 [DanC]
- I'm at http://belo.stanford.edu:8080/iwbrowser/WffBrowser?url=http://www.ksl.stanford.edu/software/IW/sample/TS/IW1.daml
- 18:29:46 [marja]
- I see what danc is seeing
- 18:29:59 [DanC]
- premises to ultimate conclusions I see is "For all values of ?inst: ?inst is a seafood when ?inst is a crab." and "Tony is a crab."
- 18:30:11 [DeborahMc]
- http://belo.stanford.edu:8080/iwbrowser/index.jsp
- 18:30:59 [DanC]
- now I'm at http://belo.stanford.edu:8080/iwbrowser/WffBrowser?url=http://www.ksl.stanford.edu/software/IW/tmp/IW1.daml
- 18:31:43 [DanC]
- eek! same address as before, but different stuff on the screen!
- 18:33:21 [DanC]
- timbl: this is an example of why web architecture says URIs should be unambiguous.
- 18:34:05 [RalphS]
- RalphS has joined #sw-team
- 18:34:09 [ericP-scribe]
- debm: we want feedback on the explainations (rules product)
- 18:37:58 [DanC]
- in N3, { :shellfish s:subClassOf :seafood } is log:includes of [ is log:semantics of <http://www.ksl.stanford.edu/DAML/query-answer/wines-short2.daml>]
- 18:38:59 [sandro]
- deb: side conditions are more premises that we dont anticipate you'd want to see in the proof
- 18:40:05 [DanC]
- patterns of sentences... that resulted in wtr in earlier KIF versions. i.e. quoting ;-) but I gather the CL folks threw out quoting recently
- 18:40:35 [sandro]
- jos: how can I see what the bindings are?
- 18:40:47 [sandro]
- paulo: there;s a supporting document that's not available now
- 18:40:56 [sandro]
- deb: click on "v" box
- 18:41:08 [DanC]
- [[ ?x = |http://www.ksl.stanford.edu/DAML/query-answer/wines-short2.daml#|::|SEAFOOD| ]]
- 18:41:56 [ericP-scribe]
- (the "v" box is after "Tony is a seafood.")
- 18:42:02 [sandro]
- paulo: most reasoners seem to lose bindings when they dump a proof
- 18:42:16 [sandro]
- ack DanC
- 18:42:16 [Zakim]
- DanC, you wanted to check that we're all on the same page and to note including source info in the logic
- 18:42:51 [ericP-scribe]
- danc: having source info off to the side... this is a critical part of the proof. in N3 it's part of the formula.
- 18:42:56 [DanC]
- in N3, { :shellfish s:subClassOf :seafood } is log:includes of [ is log:semantics of <http://www.ksl.stanford.edu/DAML/query-answer/wines-short2.daml>]
- 18:43:08 [ericP-scribe]
- debm: some folks just dump (provenence) on the floor.
- 18:43:31 [ericP-scribe]
- sandro: look at format behinde this?
- 18:43:42 [danbri]
- zakim, drop danbri
- 18:43:42 [Zakim]
- DanBri is being disconnected
- 18:43:43 [Zakim]
- -DanBri
- 18:44:04 [ericP-scribe]
- debm: first, following danc's comment, click on second explaination point.
- 18:44:38 [RalphS]
- "!" are "explanation points"
- 18:45:02 [ericP-scribe]
- paulo: a large proof will have a huge contribution from ground facts.
- 18:45:25 [ericP-scribe]
- danc: you believe this or you don't.
- 18:45:32 [ericP-scribe]
- tim: you can go back to the sources.
- 18:46:13 [ericP-scribe]
- tim: presumably you gave it these documents. to make it complete, you can point back to where you gave it the documents.
- 18:46:30 [ericP-scribe]
- paulo: every step can be related to many documents.
- 18:46:55 [ericP-scribe]
- debm: in this particular case, all came from wine-test.
- 18:47:18 [ericP-scribe]
- tim: you told it where where to get these documents.
- 18:47:30 [ericP-scribe]
- ... the eventual proof will be "you told me so".
- 18:47:49 [ericP-scribe]
- ... interesting to know what docs were used.
- 18:47:57 [sandro]
- tim: "this is very nice" :-)
- 18:48:13 [ericP-scribe]
- ... eg, filling out a tax form and discovering it used a documment you thought was private/personal...
- 18:48:31 [DeborahMc]
- one take away we will add is a summary of all documents used (along with the listing we already have of all ground axioms used)
- 18:48:56 [ericP-scribe]
- paulo: ... creating an annotation mechanism for annotating the sources of ground axioms of your proof.
- 18:49:09 [ericP-scribe]
- sandro: maybe you can point that out in the format.
- 18:50:01 [drewmcd]
- drewmcd has joined #sw-team
- 18:50:15 [ericP-scribe]
- debm: the document icon will take you to the source document of the current node
- 18:50:28 [ericP-scribe]
- paulo: each node of the proof is a separate file.
- 18:50:39 [sandro]
- Welcome, Drew. Are you going to call in?
- 18:50:45 [ericP-scribe]
- ... could combine them; there's a tradeoff
- 18:51:00 [marja]
- visual comments: why are the right side lines and left side lines not at the same level? and some more space between the sides would be nice
- 18:51:01 [drewmcd]
- I don't think I'm going to call in
- 18:51:30 [ericP-scribe]
- expected you might want to see the whole inference web spec, but it may be easier via this example.
- 18:52:01 [ericP-scribe]
- debm: we are now looking at the node "the type of tony is seafood."
- 18:52:39 [ericP-scribe]
- sandro: not quite following. these files have a single proof step. are there others that have combinded proof steps?
- 18:52:51 [DanC]
- (timecheck... agenda calls for us to move on to the 2nd C at xx:45)
- 18:52:53 [DeborahMc]
- http://www.ksl.stanford.edu/software/IW/tmp/ is a listing of all of the files used
- 18:52:57 [ericP-scribe]
- paulo: not now. we have generated some cases but not using them now.
- 18:53:14 [Nobu0]
- Nobu0 has joined #sw-team
- 18:53:31 [RalphS]
- [when I follow the source link (document icon), I get an RDF/XML source node that doesn't fit in the fixed-sized browser popup window and the window (is presumably forced) to have no scrollbars]
- 18:53:34 [ericP-scribe]
- paulo: one thing you can do is load a proof as an ontology [rdf graph].
- 18:54:14 [ericP-scribe]
- debm: http://www.ksl.stanford.edu/software/IW/tmp/ has all the files that lead to tony is a seafood.
- 18:54:28 [ericP-scribe]
- ... you'll see the 13 pieces of the proof.
- 18:54:35 [sandro]
- danc, understood.
- 18:54:44 [ericP-scribe]
- ... now looking at IW1.daml
- 18:55:05 [sandro]
- http://www.ksl.stanford.edu/software/IW/tmp/IW1.daml
- 18:55:09 [ericP-scribe]
- sandro: could you walk us through it more closely?
- 18:55:27 [ericP-scribe]
- paulo: releven part is the iw:WFF
- 18:55:28 [RalphS]
- [I confirm that the first part of http://www.ksl.stanford.edu/software/IW/tmp/IW1.daml appears to match the thing I can't scroll in the source node popup]
- 18:56:34 [ericP-scribe]
- (type Tony SEAFOOD) is a label from the for a node in this proof.
- 18:56:49 [ericP-scribe]
- debm: if you wanted to check this proof, you'd get that string out.
- 18:57:05 [JosD]
- q+
- 18:57:30 [ericP-scribe]
- q+ to ask if iw:KIF denotes <iw:Statement> contents are parsable as KIF
- 18:57:58 [ericP-scribe]
- tim: proof is no longer an RDF graph, it has embedded KIF.
- 18:58:23 [ericP-scribe]
- paulo: it's difficult to get as good as KIF
- 18:58:35 [ericP-scribe]
- tim: why use any RDF at all?
- 18:58:49 [ericP-scribe]
- debm: want to be compatable with RDF and web.
- 18:59:05 [sandro]
- zakim, q?
- 18:59:05 [Zakim]
- I see JosD, ericP-scribe on the speaker queue
- 18:59:20 [ericP-scribe]
- danc: would make sense to do it all in KIF or RDF
- 18:59:45 [DanC]
- timbl notes connection between (type Tony SEAFOOD) and <iw:Term>|http://www.ksl.stanford.edu/DAML/query-answer/wines-short2.daml#|::|SEAFOOD|</iw:Term>
- 19:00:02 [ericP-scribe]
- tim: when i parse that KIF, i have to take it on faith that their aren't two SEAFOODs.
- 19:00:12 [ericP-scribe]
- debm: this is the one without namespaces.
- 19:00:59 [DanC]
- (I've got this paper that sandro mentioned, from Maimi, entitled "Inference Web: Portable Explanations for the Web"; I wonder if I can share it with those present via http URI)
- 19:01:31 [ericP-scribe]
- ericP: you can do namespaces as a pre-processor
- 19:01:31 [sandro]
- ack JosD
- 19:01:54 [sandro]
- JosD: Tony, SEAFOOD, etc should be URIs or qnames or something
- 19:02:25 [ericP-scribe]
- paulo: we can support that. it's just a hack. could have it by afternoon^H^H^H^H^Hevening
- 19:02:41 [DanC]
- ah... http://www.ksl.stanford.edu/people/pp/papers/McGuinness_SSS_2003.pdf has pretty much the same content as the paper Sandro mentioned, with the WFF spelled out as an s-expression in RDF.
- 19:02:51 [ericP-scribe]
- jos: see no convention for labeling rules
- 19:02:55 [DanC]
- from InferenceWeb publications http://www.ksl.stanford.edu/software/IW/papers/
- 19:03:09 [ericP-scribe]
- paulo: there is no inference step related to IW1.
- 19:03:16 [ericP-scribe]
- ... but you can in this format.
- 19:03:34 [ericP-scribe]
- ... each inference step is the application of one inference rule.
- 19:04:40 [ericP-scribe]
- ... can have link to other antecedents of this rule
- 19:05:01 [sandro]
- "hasVariableMapping" == "binding"
- 19:05:33 [ericP-scribe]
- debm: node lable (kif), consequent of inf step using GMP using inf engine JTP.
- 19:05:43 [ericP-scribe]
- ... two antecedents IW2 and IW6
- 19:05:55 [ericP-scribe]
- ... variable binding ?x := SEAFOOD.
- 19:08:04 [sandro]
- somewhat updated http://www.w3.org/2003/04/29/swad/
- 19:08:16 [JosD]
- q+
- 19:08:47 [sandro]
- ack JosD
- 19:08:47 [DanC]
- ack JosD
- 19:08:52 [ericP-scribe]
- ericP: connection between "SEAFOOD" in "(type Tony SEAFOOD)" and ?x iw:Term "...SEAFOOD".
- 19:09:15 [ericP-scribe]
- jos: what is the label of the implication itself?
- 19:09:19 [DanC]
- <iw:InferenceRule rdf:about='http://www.ksl.stanford.edu/software/IW/registry/IR/GMP.daml'/>
- 19:09:47 [ericP-scribe]
- debm: it's in the inference step.
- 19:09:53 [ericP-scribe]
- ... you have the rule: GMP
- 19:10:04 [ericP-scribe]
- ... antecedents: IW2, IW6
- 19:10:10 [ericP-scribe]
- ... and the consequence.
- 19:10:58 [ericP-scribe]
- debm: either IW2 or IW6 is the implication
- 19:11:13 [DanC]
- i.e.
- 19:11:14 [DanC]
- <iw:WFF rdf:about='http://www.ksl.stanford.edu/software/IW/tmp/IW2.daml'/>
- 19:11:15 [ericP-scribe]
- sandro: the inference step and the rule have the same URI
- 19:11:59 [ericP-scribe]
- ... if you separate them, you can graph merge these files.
- 19:12:11 [DanC]
- sandro: if you give the formula in IW2 its own name, you can merge it into the proof file.
- 19:13:04 [ericP-scribe]
- jos: came to the conclusion that "Tony is a seafood", but where is it written?
- 19:13:17 [ericP-scribe]
- danc: how is that not http://www.ksl.stanford.edu/software/IW/tmp/IW2.daml ?
- 19:13:28 [DanC]
- <iw:Statement>(<= (type Tony ?x) (subClassOf CRAB ?x))</iw:Statement>
- 19:14:18 [Nobu]
- Nobu has joined #sw-team
- 19:14:56 [tim-mit]
- http://www.w3.org/2000/10/swap/test/reason/t4.proof <- comparable cwm proof. Use of {} in place of embedding kif in rdf,... variable names are uris,....parsing,....
- 19:15:21 [ericP-scribe]
- debm: we could record how the statement was proved, not just the value of it.
- 19:15:31 [ericP-scribe]
- ... seems hard; what's the value?
- 19:15:56 [ericP-scribe]
- tim: suppose we com across this implication and we want to use it elsewhere.
- 19:16:12 [ericP-scribe]
- ... you refer to it via the proof step that led to it.
- 19:16:49 [ericP-scribe]
- sandro: hasAntecedent points to the file
- 19:17:07 [ericP-scribe]
- tim: semantics to the extra step.
- 19:17:21 [Nobu]
- Nobu has joined #sw-team
- 19:17:21 [ericP-scribe]
- sandro: need about="" to keep filename
- 19:17:49 [DanC]
- <WFF rdf:about="#thisWF"> would be better, yes
- 19:17:52 [tim-mit]
- about="#wff"
- 19:18:23 [tim-mit]
- <IW2.daml#wff>
- 19:19:13 [ericP-scribe]
- debM: small addition to PPF, which is fine.
- 19:19:30 [DeborahMc]
- we do not see it
- 19:20:22 [Zakim]
- -Danny
- 19:20:35 [JosD]
- bye DanC
- 19:20:54 [ericP-scribe]
- tim compares http://www.w3.org/2000/10/swap/test/reason/t4.proof to PPF...
- 19:20:59 [DanC]
- not, that was Danny Weitzner excusing himself; I'm still here.
- 19:21:35 [Nobu]
- Nobu has joined #sw-team
- 19:21:46 [ericP-scribe]
- tim: one question, how often do you give a copy of the conclusion?
- 19:21:56 [DanC]
- I think "erasure" is a more traditional term than "extraction". hmm... part-of-speech mismatch.
- 19:22:06 [ericP-scribe]
- ... ... denoted by "gives" here.
- 19:22:24 [ericP-scribe]
- ... use/mention bug in non-quoted variables
- 19:23:13 [ericP-scribe]
- jos: i think you can perfectly use it if you say that verable=that value as part of the explaination.
- 19:23:25 [DanC]
- jos: another option is to say <t4.n3#x> = <c>
- 19:23:39 [DanC]
- i.e. really use the variable, rather than mentioning it in a binding
- 19:24:33 [DanC]
- timbl: problem with referential transparency... if x is bound elsewhere to d, we conclude c=d, no? [DanC wonders...]
- 19:24:34 [drewmcd]
- drewmcd has joined #sw-team
- 19:24:52 [JosD]
- http://www.agfa.com/w3c/euler/gedcom-proof.n3
- 19:25:35 [ericP-scribe]
- sandro: tim, you have a good sense of what inference is doing. do you think cwm could output their files.
- 19:25:48 [JosD]
- what I meant was eg {?C2 = :Rita. ?F = :dp. ?C2 gc:childIn ?F} => {:Rita gc:childIn :dp}.
- 19:26:02 [DanC]
- doing the design in KIF and putting the <>s on later would be easier, for me
- 19:26:22 [ericP-scribe]
- tim: problem is the mixture of KIF.
- 19:26:36 [ericP-scribe]
- ... the KIF can be arbitrarily complex
- 19:26:49 [DanC]
- KIF syntax is *much* simpler than s-expr-in-RDF syntax.
- 19:28:59 [DanC]
- I've implemented URI namespaces in KIF
- 19:29:02 [ericP-scribe]
- sandro: KIF is easy to parse.
- 19:29:21 [ericP-scribe]
- ... but we need a standard for what URIs mean in KIF.
- 19:29:36 [ericP-scribe]
- debm: namespace issue is not a big one for KIF.
- 19:29:45 [DanC]
- http://www.w3.org/2000/10/swap/kifExpr.html
- 19:30:10 [tim-mit]
- wine<pickyour punctuation symbol>SEAFOOD
- 19:30:33 [DanC]
- http://www.w3.org/2000/10/swap/KIFSink.py
- 19:30:57 [ericP-scribe]
- sandro: the iw:KIF element is a crucial part of this proof.
- 19:31:21 [drewmcd]
- What would a "standard for what URIs mean" look like?
- 19:31:30 [ericP-scribe]
- debm: it would be easy to add <iw:n3>
- 19:31:47 [ericP-scribe]
- debm: ansi-KIF and ansi-stamped
- 19:31:54 [ericP-scribe]
- danc: is it the bible?
- 19:32:03 [Nobu]
- Nobu has joined #sw-team
- 19:32:07 [ericP-scribe]
- debm: nah. we use ansi kiff.
- 19:32:28 [tim-mit]
- debm: we can read either
- 19:32:35 [sandro]
- Deb: a lot of people still using KIF 3.0 because ansi kif dropped some things people cared about
- 19:32:46 [ericP-scribe]
- ... some use kif3.0 'cause ansi dropped a couple things that people care about
- 19:32:49 [DanC]
- ack DanC
- 19:32:49 [Zakim]
- DanC, you wanted to talk about registration and verification
- 19:33:11 [ericP-scribe]
- dabc: stronger form of weakly queue is the one i care about.
- 19:33:21 [RalphS]
- s/weakly queue/wtr/
- 19:33:27 [RalphS]
- (weakly true)
- 19:33:33 [ericP-scribe]
- tim: brackets are for putting n3 in n3
- 19:33:35 [tim-mit]
- deb, are you there?
- 19:33:38 [sandro]
- DEB -- we cant hear ytou
- 19:33:44 [DanC]
- s/one I care about/one that seems most relevant/
- 19:33:51 [tim-mit]
- zakim, who is here?
- 19:33:51 [Zakim]
- On the phone I see DanC, Marja, JosD, Stanford/KSL, TimsOffice, EMiller
- 19:33:53 [Zakim]
- Stanford/KSL has DebM, Paulo
- 19:33:54 [Zakim]
- TimsOffice has TimBL, Sandro, Ryan, EricP, RalphS
- 19:33:55 [Zakim]
- On IRC I see Nobu, drewmcd, RalphS, DeborahMc, tim-mit, JosD, Danny-DC, ryan-lap, reagleMIT, RRSAgent, Zakim, marja, danbri, em, dom, DanC, sandro, ericP-scribe, Ralph
- 19:34:15 [ericP-scribe]
- ... want to use n3 on the inside and the outside.
- 19:34:44 [ericP-scribe]
- danc: cheaper to get rid of the points and use KIF.
- 19:35:43 [ericP-scribe]
- tim: declare the ns at the top and quote the document
- 19:35:47 [DanC]
- I prefer uniform, self-similar syntax to any particular syntax. i.e. all-RDF or all KIF or all N3 is preferable to XML/KIF or KIF/N3 mix.
- 19:35:49 [drewmcd]
- drewmcd has joined #sw-team
- 19:36:00 [sandro]
- tim: It would be great to have a std way to put URIs into and out of KIF.
- 19:36:12 [sandro]
- tim: If we had that, write your proof language in KIF.
- 19:36:24 [sandro]
- ...: If you need the power of KIF, you need the power of KIF.
- 19:36:43 [sandro]
- ...: the mixture of RDF/XML and KIF seems messy.
- 19:37:17 [em]
- q+
- 19:37:18 [sandro]
- Deb: followup on NS in KIF
- 19:37:20 [ericP-scribe]
- debm: action item for a subset of us to have conversation on ns in kiff?
- 19:37:40 [ericP-scribe]
- danc: that's not my biggest issue with this PPF.
- 19:37:48 [em]
- q-
- 19:37:58 [sandro]
- MAJOR ISSUE: KIF
- 19:38:23 [Nobu]
- Nobu has joined #sw-team
- 19:38:23 [ericP-scribe]
- ... point of a proof lang from my angle is that a verifier is a small bit of code.
- 19:38:39 [ericP-scribe]
- ... that code is the proof checker.
- 19:39:10 [ericP-scribe]
- ... when you install it locally, you have it sign its output.
- 19:39:25 [Nobu]
- Nobu has joined #sw-team
- 19:40:14 [ericP-scribe]
- ... i want my locally installation of JTP to sign it's output.
- 19:40:38 [ericP-scribe]
- ... and tell proofchecker the corresponsing pub key.
- 19:41:01 [ericP-scribe]
- ... one way: everyone signs up with stanford.
- 19:41:26 [ericP-scribe]
- ... other way: trust the signatures you know.
- 19:41:48 [ericP-scribe]
- debm: expect both to happn
- 19:42:23 [ericP-scribe]
- sandro: do you wnat proof format to distinguish between ...
- 19:42:30 [DeborahMc]
- signature checking as a proof step is an extension we will add
- 19:42:37 [sandro]
- DanC: I just want signature verification to be a proof step
- 19:42:45 [ericP-scribe]
- tim: other way to support trust is to look it up on the web.
- 19:43:05 [ericP-scribe]
- ... anything that document A references is true.
- 19:43:28 [ericP-scribe]
- danc: that's an innovation in theorum provers.
- 19:43:39 [ericP-scribe]
- tim: do you have arithmatics built in?
- 19:43:51 [ericP-scribe]
- debm: minimally.
- 19:44:41 [ericP-scribe]
- danc: they have no builtins, just registered theorums and provers.
- 19:45:01 [ericP-scribe]
- ... a certain number of these have to be blessed in a special way.
- 19:45:31 [ericP-scribe]
- ... the proof check must know about a certain nnumber of inference rules.
- 19:46:05 [ericP-scribe]
- debm: i think we'll have to do some registration for internal proof checking.
- 19:46:14 [ericP-scribe]
- tim: crypto rules are particularly interesting
- 19:46:27 [ericP-scribe]
- debm: we would expect to partner on this.
- 19:46:44 [ericP-scribe]
- danc: have you looked at Felton's work?
- 19:46:47 [ericP-scribe]
- jos: i have
- 19:47:02 [ericP-scribe]
- danc: 'cause they are working on caching and speeding up
- 19:47:23 [ericP-scribe]
- jos: putting burden on client makes job more modular.
- 19:47:48 [DanC]
- burden on the client to generate the proof
- 19:47:57 [DanC]
- Felten and company on PCA http://www.cs.princeton.edu/sip/projects/pca/
- 19:48:47 [tim-mit]
- Absolutely
- 19:49:00 [ericP-scribe]
- jos: my making refrences to pieces that you would otherwise not know, you are proving ...
- 19:49:28 [ericP-scribe]
- ---------------------------------
- 19:49:31 [sandro]
- sandro: what next?
- 19:49:33 [ericP-scribe]
- sandro: what happens next?
- 19:49:50 [ericP-scribe]
- debm: one thing would be a small PP extension
- 19:49:57 [DanC]
- "portable proof extension?"
- 19:50:21 [ericP-scribe]
- ... converstaion between me and someone on that side, dunno who is best
- 19:50:27 [ericP-scribe]
- tim: extension for what?
- 19:50:35 [ericP-scribe]
- debm: premise labeling
- 19:51:17 [ericP-scribe]
- debm: we need to get your infrence rules registered
- 19:51:35 [ericP-scribe]
- ... paulo didn't think it looked hard either.
- 19:51:39 [drewmcd]
- quit phone call
- 19:51:58 [ericP-scribe]
- jos: puting it in n3 is the matter of a couple hundred lines.
- 19:52:18 [ericP-scribe]
- debm: some experts on the phone were not confident.
- 19:52:44 [ericP-scribe]
- debm: what we have registered is what we display.
- 19:53:16 [ericP-scribe]
- [scribe is verrry sleepy now]
- 19:53:26 [ericP-scribe]
- [to hot in here]
- 19:54:11 [ericP-scribe]
- sandro: what about tim's point that KIF in RDF is ugly. need all round brackets or all pointy brackets.
- 19:54:25 [ericP-scribe]
- so hot that...
- 19:55:02 [ericP-scribe]
- debm: not a big prob. JTP does it all in KIF. with Jessica in the room, the problem will be solved.
- 19:55:51 [ericP-scribe]
- danc: cwm could spit out nonsense and it (IW) would say "nifty".
- 19:56:07 [ericP-scribe]
- debm: yes, today. but tomorrow, hopefully not.
- 19:56:36 [DanC]
- i.e. IW sorta assumes the registered provers produce coherent/sound/valid/whatever proofs
- 19:57:07 [sandro]
- but of course some verifier could work in parallel on the same inputs....
- 19:57:36 [ericP-scribe]
- tim: you are going to a higher level of abstraction
- 19:57:59 [ericP-scribe]
- jos: some more built-ins would be very relevent.
- 19:58:09 [ericP-scribe]
- ... cwm has close to 80 or so.
- 19:58:18 [ericP-scribe]
- ... mathematics and crypto.
- 19:58:23 [sandro]
- builtins == "procedural attachments" in RuleML-speak
- 19:58:32 [ericP-scribe]
- ... how would that be done in KIF?
- 19:59:06 [ericP-scribe]
- debm: tim: i think they'll be some standardization of these.
- 19:59:14 [ericP-scribe]
- s/^debm: //
- 19:59:44 [ericP-scribe]
- tim: for instance, the community will assume RDF can use the XML query primitives.
- 19:59:59 [tim-mit]
- http://www.w3.org/2000/10/swap/doc/CwmBuiltins
- 20:00:12 [tim-mit]
- ^ List of cwm's current built ins
- 20:00:53 [sandro]
- I'll send a pointer to www-archive and us.
- 20:00:56 [ericP-scribe]
- sandro: could send pointer to minutes to and archive, or www-rdf-logic or www-rdf-rules.
- 20:01:16 [ericP-scribe]
- debm, danc: just archive.
- 20:01:17 [sandro]
- ADJOURN
- 20:01:21 [ericP-scribe]
- [closed]
- 20:01:32 [Zakim]
- -EMiller
- 20:01:43 [Zakim]
- -DanC
- 20:01:46 [em]
- thanks folks, have to run
- 20:02:10 [Zakim]
- -JosD
- 20:02:15 [Zakim]
- -Marja
- 20:03:07 [Zakim]
- -TimsOffice
- 20:06:02 [DeborahMc]
- we hung up but are still on irc
- 20:06:30 [Zakim]
- SW_SWAD()2:00PM has ended
- 20:07:21 [RalphS]
- RalphS has left #sw-team
- 20:13:17 [Nobu]
- Nobu has joined #sw-team
- 20:24:55 [reagleMIT]
- reagleMIT has left #sw-team
- 20:25:35 [Ralph]
- rrsagent, please excuse us
- 20:25:35 [RRSAgent]
- I see no action items