<vkolovski> Date: 26 Oct 2005
<timbl> freeloader agency wiht Jim's group for a year on sem web stuff with Jim H
SC: Saw PAW stuff. thought i's listen in.
DanC: We have met
Steve: yes. I'll be at ISWC
<timbl> Scribe: Vlad
<timbl> ScribeNick: scribe
timbl: going over the action
items...
... vlad have you received a confirmation for your member
access
vlad: i've been working on
pychinko, haven't committed anything to the cwm cvs. will check
whether my account works shortly.
... i did receive a confirmation email from a member of the w3c
staff
<scribe> ACTION: Tim to get Vlad cwm CVS access (9/28: Vlad appplies for W3C passwrord) (Request sent to systeam 2005/10/12) [DONE]
timbl: round the table..
danK: finished the supprot for n3 in rdflib
<timbl> rdf://rdflib.net/
<timbl> http://www.rdflib.net/
danK: Now pychinko converted to
the new rdflib. The forked verison is cut over
... Now also, the dictionalry relating bnodes to formulae has
gone.
... This correspnds more closely to the cwm architecture.
... Also found a bug with vartiables..... in pychnko you have a
dependencey graph., and do a topological sort on it, and then
build the rete network.
Vlad: Fixed bug / missing functionality regarding dependencies by adding topological sort
DanC: what about loops
Vlad: Only for builtins
<timbl> Non-builtins are put first and then the built-ins are all sorted. This is within one rule.
<vkolovski> we have nonbuiltins, then we topological sort the other alphanodes
<timbl> Paper options: 1. semantics of N3 9or a subset)
<timbl> Yarden has commited on that, Vlad can help out -- vlad
<timbl> paper on cwm as a theorm prover for the WWW, or compare it wiht PCA.
<timbl> Deadlines m kR 6th November
<timbl> Other deadline WWW is very soon too 4th Nov approx.
<timbl> The cwm paper -- a systems paper.
<timbl> Jim suggested the papers.
<scribe> ACTION: Vlad to talk to jim about the papers and their chances of getting in the conferences.
<timbl> Tim: I am happy to explain the N3logic stuff where necessary and review things as they evolve.
<timbl> Vlad: Maybe more operational semantcis, and talk to Jordan about it
<timbl> DanC: How close is pychinko to hte rein example?
<timbl> Vlad? CLose to running it without proof.
Vlad: re:paper - our preference
is a semantics paper for KR, we wil ltry to cover as much n3 as
possible
... even if we have a restrictive subset , we should be in a
good shape.
timbl: working on --why
... found out it was not making proofs of log:includes
properly
... gotto a point where it will to the tests
... was difficult to do proofs looking for bnodes
... there were problems with nested lists
<timbl> unify() is improved
<timbl> --n3=B turns bnodes into named existentitals
timbl: got that working this
morning
... i've got the rein proof generated and checked
danc: i've got the proxy almost ready to call cwm.
<DanC> http://www.policyawareweb.org/2005/demo/pfproxy.py
danc: do we have a demo documentation page?
(discussion about the location of the proof)
<DanC> Examples for Rein Policy Framework
<DanC> not a demo doc per se, but pretty close
<DanC> tim: I haven't run all the tests; should I check my code in anyway?
timbl: will check in the proofe generation and checking code in the main cvs
<DanC> DanC: you've run some? then yes.
<timbl> http://www.w3.org/2000/10/swap/test/reason/judy-proof.n3
<DanC> vkolovski, have you tried pfproxy?
<DanC> eikeon_, you tried pfproxy a little, yes?
<vkolovski> i haven't
<eikeon_> DanC, I've looked at it. But not yet tried sending traffic through it.
<timbl> @forSome :_g10,
<timbl> :_g11,
<timbl> jud:_g_L8C18 .
<timbl> :_g10 http:can-get <http://demo.policyawareweb.org/images/group.jpg> .
<timbl> :_g11 http:can-get <http://demo.policyawareweb.org/images/group.jpg> .
<timbl> judy:req rein:ans rein:Valid;
<timbl> rein:requester :_g10,
<timbl> :_g11 .
<timbl> _________________________
<DanC> Subject: [Paw-team] Re: checked in a proxy (ready to call cwm/rein)
<DanC> Date: Thu, 06 Oct 2005 11:55:31 -0500
timbl: i will run some more tests on it
<DanC> http://www.policyawareweb.org/2005/demo/pfproxy.py
danc: i have the proxy ready for the next step
timbl: the q is how it's going to take for the actual thing to run (the proof is huge)
danc: there will be all kinds of engineering for the proof
danK: i am working on the demo
and am available on #swig all the time
... also can pop up on #paw
<DanC> DK: sure, I can be on #swig and #paw for the most part from now to Fri AM
danc: have you thought about running check.py?
danK -> DK
DK: we will need the latest cwm from the cvs, right
<timbl> cwm.py,v 1.178 2005/10/24 16:58:38 timbl Exp
<timbl> Repository revision: 1.17 /w3ccvs/WWW/2000/10/swap/check.py,v
timbl: round table updates - don
<scribe> done
<DanC> victory! ACTION: TimBL to work on --why toward log:includes, log:conclusion for rein
<DanC> ACTION: TimBL to work on --why toward log:includes, log:conclusion for rein [DONE]
<timbl> --why --base=foo
<timbl> <>
timbl: PML ...
... DanC, you wanted to discuss that
Danc: I've been putting off that stuff because of the demo
<DanC> vicotyr! "RDFlib fork - where were we?" forks are joined
<DanC> yup, in N3 you can say @forSome <http://www.w3.org/foo#bar>.
DK: discussing exivar/bnodes in
n3
... should we have two node types for bnodes and existential
vars?
<timbl> [] color green.
<vkolovski> pychinko does not throw away the list triples, they'are used in matching
<vkolovski> it does convert them to list objects,though, which are used in builtins
<DanC> consider: @forsome <x>. <x> <y> <z>.
timbl: in cwm, all variables have a URI
<DanC> N3 is wierd this way.
<DanC> [] <y> <z>.
<DanC> consider: @forsome <x>. <x> <y> <z>.
<DanC> ok to read that in and spit it out thusly: [] <y> <z>.
<DanC> tbl: no...
<DanC> true or false: { @forsome <x>. <x> <y> <z>. } log:includes { [] <y> <z> } ?
<DanC> true or false: { _:foo <y> <z>. } log:includes { [] <y> <z> } ?
timbl: if the existential and
universial variables have explicit names, you assume it was
done for a reason and you dont want to rename them
... cwm does not store the bnode ids
danc: it seems that all
restrictions apply to exivars, univars and node_ids
... is there a specific reason for "forgetting" bnode ids
timbl: if you merge two files, you have to do renaming of existentials, but not universals
<DanC> { @forSome <x>. <x> <y> <z>. } log:includes { [] <y> <z> } # cwm says yes
timbl: unification engine in
query.py
... unify() method works on nested formulae, but it's too slow
to work on anything serious
... i see that we can benefit from the rete engine
vlad: next steps - actually taking the rete engine and integrating it in cwm
<DanC> 9 Oct conflicts with ISWC, I gather.
<DanC> 4pET Fr 4 Oct? danc, timbl can do... DK at risk
RESOLUTION: to meet again 4pET Fr 4 Oct, DK at risk.
RESOLVED to cancel 9 Oct in favor of ISWC
<DanC> on 16 Nov DanC is speaking at XML2005
<DanC> 23 Nov 2:30pET seems OK
<DanC> tx, vlad, for offering to send me a log
<timbl> I'll be around for a bit probably though if you want to keep the momentum. @@postscript
