paw cwm-dev

15 Feb 2006

Attendees

Present
TimBL, DanK, Vlad, DanC, Yosi
Regrets
Jordan
Chair
TimBL
Scribe
DanC

Contents


 

 

Convene, take roll, review records and agenda

<timbl> on the phone I hear vlad, eikeon

<timbl> MEETING STARTS

<DanC_> I'd like to move PML stuff (currently #7) up, in time, if not on that page

timbl: agenda + n3.n3 status

vlad's been working on/near 6 integrating pychinko and cwm

(looking ahead 2 week, timbl and DanC aren't available 1 Mar)

Yosi: tue/thu 11-1 is my most available time
... this meeting interrupts an otherwise useful time

(I can't fit a meeting on tue. thu I can do, maybe)

TimBL: Thu 11aET then?

DanC: starting 9 Mar?

RESOLUTION: to meet Thu 9Mar 11aET for ~90min then

integrating pychinko and cwm

<DanC_> connolly@dirk:~/w3ccvs/WWW/2000/10/swap$ cvs update -d pycwmko

<DanC_> cvs update: nothing known about pycwmko

<DanC_> connolly@dirk:~/w3ccvs/WWW/2000/10/swap$ cvs update pycwmko.py

<timbl> pycwmko.py

<timbl> --pythink

VK: it groks a limited subset of N#
... it groks a limited subset of N3

<yosi_s> cwm test/rules-simple.n3

<yosi_s> --pythink

<vlatko> @prefix : <http://cwmTest/>.

<vlatko> @prefix log: <http://www.w3.org/2000/10/swap/log#>.

<vlatko> @prefix math: <http://www.w3.org/2000/10/swap/math#> .

<vlatko> {?x :parent ?y} log:implies {?y :child ?x}.

<vlatko> <http://cwmTest/a5776> <http://cwmTest/parent> <http://cwmTest/b18684>.

<timbl> cwm rules-simple.n3 --pythink

<DanC_> connolly@dirk:~/w3ccvs/WWW/2000/10/swap$ PYTHONPATH=~/paw-mindswap/pychinko/ python cwm.py --pythink test/rules-simple.n3

<vlatko> http://svn.mindswap.org/pychinko/

<DanC_> % svn co svn+ssh://mindswap.org/home/svn/paw # per http://www.policyawareweb.org/2005site/dev.html

<DanC_> oops... that might not be where pychinko lives

<DanC_> LOG = Namespace("http://www.w3.org/2000/10/swap/log#")

<DanC_> TypeError: 'module' object is not callable

<DanC_> Updated to revision 5879.

<DanC_> Python 2.3.5 (#2, Aug 30 2005, 15:50:26)

<DanC_> from rdflib import Namespace

<timbl> pychinko/pychinko/terms.py

<DanC_> File "/home/connolly/paw-mindswap/pychinko/pychinko/terms.py", line 10, in ?

<eikeon_> http://svn.rdflib.net/trunk/

<DanC_> if p!= LOG.implies

Yosi: I'm taking a look at directPychinkoQuery , and it doesn't [???]

VK: I did some tests with 7000 [something] and it seems good enough

YS: last Aug I noticed an analog between indexed forumla and [something?] so I used it directly.

TimBL: so that's more efficient?

YS: more direct; different scaling. dunno if it's more efficient

TimBL: [... on how cwm works... beyond what I can follow ...]
... that's another way to do it: have a mode where the KB is always up-to-date with all the rules

VK: we've had to do that with our demo/portal [?]
... with 50K facts, adding a rule takes 1 second

DK: we looked at that close integration with indexed formulas, but then went for easier optimizations

TimBL: let's talk more about that sort of thing
... let's work toward a cwm release that uses a rete, and explain when it's good to use

VK: the integration isn't done... right now the rete conclusions are just printed. need to feed them back

TimBL: make sure to trap errors for stuff it doesn't handle

DanC: I'd like it to pass 5 or 10 interesting tests

VK: should I put a copy of the pychinko code in the cwm codebase?

TimBL: it's a hassle, but I think so, for usability...

DanC: the pychinko code has to get copied into the released tarball, but I don't think we _need_ to copy the version controlled stuff

<timbl> Conclusion: The TAR file will have the whole source tree in, with swap/pychinko

<timbl> ... but the CVS won't

<timbl> Aim for:

<timbl> - --pythink ? -or --rete or -run

<DanC_> --rete

make rete-test

<timbl> -- tests using --rete in the test/rete/detailed.tests

<timbl> -- rouch doc'n on constaints under which you can use --rete

DanC: I'd like to hear an update on this once a week (to paw-team and/or public-cwm-talk) for the next 3 weeks
...

PAW demo item? How is it going? Debugging cwm --why and proof checking.

YS: found a number of bugs in entaiment checking [in check or llyn?]
... found a number of bugs in entailment checking check
... found a number of bugs in entailment checking in check.py
... some apects of rein really push the limits
... I got LK to change some of those

<timbl> YS: The paw test cases seemed to be very difficule to etst n3entailment.

<DanC_> [oh? where did LK make changes?]

<timbl> YS: There are some very obscure cases left wheer it will incorrectly conclude entailment.

<yosi_s> http://groups.csail.mit.edu/dig/2005/09/rein/examples/troop42-policy.n3

<timbl> cwm --rdf http://dig.csail.mit.edu/2005/09/rein/examples/judy-req.rdf --n3 http://dig.csail.mit.edu/2005/09/rein/engine.n3 --think --filter="http://dig.csail.mit.edu/2005/09/rein/filter.n3" --why

<timbl> is in the agenda page.

<timbl> Is that still the test case?

<DanC_> http://groups.csail.mit.edu/dig/2005/09/rein/bfilter.n3

<DanC_> http://groups.csail.mit.edu/dig/2005/09/rein/bengine.n3

<DanC_> connolly@dirk:~/paw/www/2005/demo$

<DanC_> how to check it out is documented in http://www.policyawareweb.org/2005site/dev.html

<DanC_> # % svn co svn+ssh://mindswap.org/home/svn/paw

YS: I've got it down to 3 minutes; I'm somewhat content with that
... I've been working on log:supports; I think I have that working

DanC: do you have a log:supports test?

YS: in progress

<DanC_> (getting TBL's key in the paw svn repo...)

<timbl> includes/conclusion.n3

<timbl> includes/conjunction.n3

<timbl> includes/decision.n3

<timbl> includes/loop.n3

<timbl> reason/engine.n3

<timbl> reason/judy-proof.n3

<timbl> reason/paw-proof.n3

<timbl> are files which have log:conclusion and log:includes

<timbl> under swap/test

cwm/Inferenceweb integration

<DanC_> log of my struggles with to-pml.n3 and such last night

<DanC_> my conversion to pml of a webarch versioning proof

<DanC_> a pml proof deely

<timbl> http://dig.csail.mit.edu/2005/ajar/ajaw/tab

<yosi_s> http://www.w3.org/2000/10/swap/test/supports/simple.n3

<timbl> about:config

<timbl> http://inferenceweb.stanford.edu/registry/DPR/GMP.owl#GMP

<ronwalf> timbl: Are you using the same private keys as you were back in August?

<DanC_> From: Pinheiro da Silva, Paulo <paulo@utep.edu>

<DanC_> To: Dan Connolly <connolly@w3.org>

<DanC_> Subject: RE: cwm inferenceweb registration

<DanC_> Date: Wed, 11 Jan 2006 16:44:22 -0700 (17:44 CST)

N3 semantics, papers

VK: as discussed at our ftf meeting in Jan, we're submitting to something else

<scribe> ACTION: Lalana be chief writer on an N3 logic aimed at AAAI's track on AI & the Web. [WITHDRAWN]

<vlatko> Journal of Theory and Practice of Logic Programming (TPLP) announces a special issue on Logic Programming and the Web

<vlatko> "Submission of papers: 2 May 2006"

<timbl> The ACTION: YK to schedule a meeting with TimBL and Bijan to discuss n3 semantics COMPLETED by Jim coming up and talking about it. Chat with Bijan will have to be some other time, not on critical path.

<scribe> ACTION: YK to schedule a meeting with TimBL and Bijan to discuss n3 semantics [DONE]

<scribe> ACTION: ACTION to Vlad chief writer on the cwm systems paper for the journal of web semantics [WITHDRAWN]

<scribe> ACTION: VK to submit cwm systems paper to Journal of Theory and Practice of Logic Programming (TPLP) announces a special issue on Logic Programming and the Web

VK: LK is submitting to this same journal

<scribe> ACTION: LK to submit N3 semantics paper to Journal of Theory and Practice of Logic Programming (TPLP) special issue on Logic Programming and the Web

"Submission of papers: 2 May 2006"

VK: we'll hear back in Oct

<vlatko> # Notification of acceptance: 19 Oct 2006

<vlatko> # Revised version of papers: 11 Feb 2007

<vlatko> # Publication: approx. end 2007

N3 changes and n3.n3

TimBL: there's a big qname regex

DanC: an sparql.n3

and bnf2turtle

another time

<DanC_> bnf2turtle -- write a turtle version of an EBNF grammar

<DanC_> bnf2turtle.py

Summary of Action Items

[NEW] ACTION: LK to submit N3 semantics paper to Journal of Theory and Practice of Logic Programming (TPLP) special issue on Logic Programming and the Web
[NEW] ACTION: VK to submit cwm systems paper to Journal of Theory and Practice of Logic Programming (TPLP) announces a special issue on Logic Programming and the Web
 
[DONE] ACTION: YK to schedule a meeting with TimBL and Bijan to discuss n3 semantics
 
[DROPPED] ACTION: ACTION to Vlad chief writer on the cwm systems paper for the journal of web semantics
[DROPPED] ACTION: Lalana be chief writer on an N3 logic aimed at AAAI's track on AI & the Web.
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2005/08/16 15:12:03 $