Policy Aware Web (PAW) Project Meeting

23 Aug 2005

This record is partially formalized using hcalendar and GRDDL. The glean-hcal transformation extracts data from this document. If you don't have a GRDDL client installed, you might be able to use the GRDDL service. The actions and event-related decisions are available in .ics and RDF/.xcal formats.


vkolovski/VK, jhendler/JH, eikeon/DK, dwood/DW, yosi/YS, Danny/DJW, TimBL, DLM, DanC. where are those photos?

Karen/KWW and Gerry joined for some parts.

for the CWM/PML/Inference Web Integration discussion, paulo, li_, and TimFinin joined remotely.

DanC, DW


Introductions and Agenda Bashing

The proposed agenda, [final revision]PAW FTF agenda of August 22, 2005 6:49:12 PM EDT, was largely accepted. DanC asked that for some of the tasks, we start by trying to actually do them; after 15 minutes or so, we would have a good feel for how long they would take.

The TAMI overview was postponed and eventually seemed to be subsumed by various parts of the discussion.

Policy Language overview and literature review (Lalana and Vlad)


<scribe> DONE: Vlad to with Lalana to add references to these slides and post on PAW page

Vlad: starting with PCA, as it seems closest to our work...

DanC: how do PCA fact servers relate to SPARQL servers?

DavidW: good question

Vlad: not a lot of documentation of the fact servers; I looked at the code and didn't see a clear design

JimH: do they serve general database queries?

Vlad: no, limited in ways [that are don't seem to be easy to characterize]

Gerry: why higher order?
... what features?

Vlad: quantification over formulas

DanC: I've never seen them justify why higher-order; their background is in this ELF stuff, and it seems to work for auth, so they went for it

Vlad: all [rules?] have to be typed before sending to server...

<DanC> [I understand that they've done a lot on making the typing implicit in order to address the "they blow up" issue]

slide PCA Architecture

[in the sense of software architecture]

> 17 Mar 2000 notes on finding PCA

<RalphS> [re: how does the client know what fact servers might be acceptable to a server -- something akin to a Web service description might work; e.g. the server doesn't need to be (the only one) to hint how to construct acceptable proofs -- anyone might do so]

<timbl> Where is the negation in that?

<DanC> I think "and I believe joe and fred about who's in what class" is an implicit premise too

<DavidW> Surely the resource server can define in its policy which sources of authentication it will accept?

<DanC> (I keep hearing JimH claim that what we're doing is different from PCA in that PCA has some limitations. As far as I know, the difference are all in the syntax-engineering realm.)

<jhendler> DanC - I think that is the key question - I don't think you are right, I think we are assuming things of PCA that may or may not be there....

TimBL: what's different is that PCA doesn't talk about documents
... well, it grounds things in keys...

DJW: we don't to just ground things in keys...

<jhendler> another difference appears to be this stuff about requiring the client to DEFINE the sublanguages that are decidable subsets of the HOL

<DavidW> Isn't grounding in keys equivalent to grounding in *some* documents?

slide PeerTrust Motivation [@@check]

<lkagal> > PeerTrust

Vlad: a difference from PCA is incremental disclosure; in PCA, the whole policy is sent, but in peertrust, just parts, and trust is built thru negotiation

lkagal, which of the N papers there is/are most relevant?

<lkagal> > an interesting PeerTrust overview paper

<DanC> [this literature review discussion has a _very_ different feel when you haven't read the papers, doesn't it?]

Vlad: PeerTrust assumes closed-world semantics. [hmm.]

slide PeerTrust Language Example

much discussion of: student(X) @ <- student(X) @ UMD @ X

yes, this @ syntax is discussed in that overview paper

<DanC> (checking to see if it cites a Flora2 paper, and I can't tell)


> TR-638-01 PCA

<Zakim> DanC_lap, you wanted to ask Gerry about "Axioms of the Core Logic" on p16 of TR-638-01

<DanC> can I get Zakim to track attendance of a meeting with no phone participants?

<RalphS> perhaps not

JimH: PeerTrust is in the same eu-funding network as Flora2. Vlad: REWERSE

slide: WS-Policy Example

TimBL: does "ExactlyOne" mean that you have proove that the others are false?

Vlad: [not clear]
... our mapping from WS-Policy to OWL is implemented in XSLT

Vlad shows OWL debugging features applied to inconsistent WS-Policy

slide: SAML and XACML

DanC: is it like Kerberosv4, with paths of length 2, each covered by shared key? or like V5 that has delegation using public key?

Vlad: SAML allows longer paths, but doens't necessarily use public key; security weakness

(ugh... urn:oasis:... . how to make the point that they're avoiding an important responsibility by hiding in urn: space and not servicing http: GET requests)

<Zakim> DanC_lap, you wanted to note that it seems to have no stronger security guarantees than OpenID, and lots more complexity/hassle and to ask to what extent SAML/XACML is

Vlad: no, we didn't try some SAML/XACML code

DJW: how is SAML/XAML not the whole solution?

Vlad: because it goes back to having all the [proof generation] on the server side

TimBL: can SAML queries be expressed in SPARQL?

several: perhaps abstractly, but practically, they're different protocols.

<lkagal> SAML also allows you to say "I believe that the sky is blue" something SPARQL cannot.

<DanC> SPARQL cannot?

<lkagal> I didn't know SPARQL could be used to make assertions.

<DanC> oh... SPARQL assumes that you just publish statements using HTTP, I guess.

JimH: most common reason papers are turned down is because they didn't cite relevant work
... agenda review

<DanC> I wonder if Zakim should grow infobot-style factoid service

(resuming from break)


WS-Policy, MINDLAB OWL mapping, and GRDDL

(SWOOP knows about WS-Policy and invokes XSLT... looks like GRDDL. I wonder if the WS-Policy folks would be willing to store the pointer in thier namespace document)

<scribe> ACTION: DanC to discuss WS-Policy mapping and GRDDL with Vlad

PAW Overview (Hendler)

[slides... pointer to be provided?]

<dlm> > the PAW paper jim just mentioned

<Zakim> timbl, you wanted to suggest two things, one the effcet of webizing policies that you can combine them; two the that you can look things up which largely solves Alice's problem.

> Paraconsistent Logic in the Stanford Encyclopedia of Philosophy

<DanC> ah... in that case, yes, re-reading that, I do think we want something paraconsistent. I think PCA's logic is paraconsistent.

<DanC> Sandro, note that this is one of the reasons I'd rather the rule language stayed positive for a year or so.

<DanC> ... and one of the reasons I asked TimBL whether he's comfortable with classical negation just the other day. I'm still somewhat surprised to find that timbl has conceded (conceding?) that point.

<DanC> ... though we clearly conceded it for OWL.


Technical Review of PAW deliverables: CMW (Tim and Yosi)

<scribe> Scribe: dwood

DanC: RDF has an XML syntax, but we haven't used it in practice
... CWM has improved automated testing, with Yosi's help
... An RC of CWM has been sent out with a SPARQL server implementation
... Proof generation and checking has been implemented and is currently running (but not guaranteed to be stable)
... (timbl demonstrated CWM working)

<DanC_lap> (odd that timbl is using <b> as a predicate in cwm; it conflicts with his preferred position on httpRange-14)

<DanC_lap> naming rules... we've talked about taking { sky color blue } <= { weather condition good} and writing it as: rule23 antecedent { sky color blue}; consequent { weather condition good}; a Truth.

<DanC_lap> I'm skeptical that the quantification details would work out.

<scribe> ScribeNick: DanC_lap

TimBL: the[sic] main difference between this [cwm --why's output] and PML is that here we quote rules literally, and in PML, they're refered to by name

discussion of rule naming PML integration...

<DanC> (I'm not writing down details because they seem straightforward to me)

TimBL: Ok, we'll hack that [i.e. naming/quoting difference]

TimBL demonstrates the round-trip: cwm --why | check | cwm

Yarden: pychinko is generating something simple, but could be adapted to the cwm syntax straightforwardly

discussion of how cwm and pychinko integrate...

TimBL: one idea was to have a version of cwm with the pychinko engine, but it needs to generate proofs too;
... so we should discuss the interface

Chris: the proof generator could just be the TMS. that's what a TMS does; it generates proofs

Yarden: pychinko doesn't do retraction

Chris: TMS doesn't do inference; it records inference

<vkolovski> (pointer to TMS?)

Chris: when we started talking ~6months ago, I had a 3 page implementation of a TMS. it's a really simple idea; I'm surprised it's considered novel.
... it's evolved from 3 pages to 10 and grown conditional proof...

DanC: how can I look at it?

Chris: I'll send it to you

timbl starts to write rules to convert cwm proof syntax to PML...

seems likely to overlap with lunch


resuming from break...

Suppes and logic.scm (Gerry)

Gerry displays logic.scm:

uncontrolled rules don't make an infinite about of stuff

<DanC> (these rules remind me of the "follow your nose" proof rules from my Philosophy 313K class at UT Austin )

JimH: I ported something like this to DAML... distributed among 3 web servers

(DanC has seen it, but neither JimH nor I remembers where it is)

Gerry: objects here are persisted into a postsgress database. connected to URIs [ in a way I didn't catch ]

> Patrick suppes

> Introduction to Logic at amazon

DLM: we used a different book: Reeves & Clark intro to C.S., and registered that stuff in the PML/InferenceWeb registry

Chris: we use URIs to relate in-memory structures to stuff on disk

Gerry: yes, the URIs we throw around start http://swiss-ai... , yes, URIs we own

DanC: just checking. some folks don't consider that obvious.


Technical Review of PAW deliverables: Pychinko (proof generating) (Yarden)

> Pychinko: Rete-based RDF friendly rule engine

YK: pychinko is a reasoner sorta like cwm that uses the state-of-the-art forward chaining algortithm, Forge's (sp?) rete
... there's an overlap with cwm... I'd rather just integrate the rete engine into cwm so that if the N3 files aren't overly expressive, it runs the rete engine

TimBL: what built-ins are covered?

YK: math and string, so far, sharing code with cwm

TimBL: we have unification... could maybe be factored out...

stuff that pychinko doesn't do:

1. any log: stuff

1.a log:includes would be 10 minutes of coding, but others...

2. lists

YK: that alone eliminates a lot of the test cases
... ... N3 semantics ... translation to transaction logic ... Keifer ... handles quoting ... less than prolog ... non-mononic ...

DJW suggests N3 semantics discussion should be postponed pending Bijan's presence...

DanC expresses some concern that this has been waiting for quite a while

DJW: the mailing list is a good way to at least record the questions

TimBL: mail to cwm-bugs has a tracker

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

<DanC> (cheers to YK and YS for copying public-cwm-talk on some of the recent collaboration)

<timbl> dlm, can you point me to a 1 or 2-page example of PL?M

"If you want to get your agent/engine to generate PML documents, your tool can already use the PML generation services for that."

> PML spec


Rein (Lalana)

[slides... pointer to be provided?]

DJW: ... policy network ... ?

answer seems to involve policy composition

Vlad: so a policy is just an N3 rules file?

LK: some policies are just RDF/OWL documents, but some [?] need to be N3 rules
... when you run cwm on all this stuff, you get a list of the request that are valid

<Zakim> DanC_lap, you wanted to suggest that Rein seems to me to be one of the answer to "How did Alice know that she should have talked to the university Registrar?"


<Danny> thx

<Danny> ack

<Zakim> timbl, you wanted to say that we have a fundamental thing to be able to do it to dereference URIs uses as identifiers, by looking up virtual or static documents on the web.

<Zakim> DanC_lap, you wanted to suggest that a rein service is either (a) on the server, in the traditional auth check place, or (b) in the client ['s computing base], helping it generate

<Zakim> jhendler, you wanted to because I cannot reconcile PAW and REIN views yet

<DanC> checkpoint: DanC seems to be in sync with DavidW: a REIN service lives in the client proxy

<dwood> DanC, The REIN service either lives in the client or is at least network-accessible to the client. I am not yet sure that the REIN service (inclusive of CWM) needs to be replicated within every client and/or proxy.

<DanC> right, acting on the client's behalf, i.e. part of the clients computing base

<dwood> DanC, yes

<DanC> (can you paste that URI into this log, lkagal ? )

<lkagal> > Rein

DW: what we don't have today is client-side proxy constructing proofs

Paw Demo (DavidW)

slide Paw Demo (showing session:hexdigest)

DavidW: we're looking at DanC's OpenID server work here too

(speaking of calendars, I use cwm in my real-life for synchronizing the calendar on my hip with my work calendar and my family calendar. cwm runs the rules very slowly, so I'm very interested in the idea of pychinko making it faster)

JimH: if you're making new use cases, put it in the use cases document

(yeah... where's "the use case document"?)

<scribe> DONE DW: link the use cases document from http://www.policyawareweb.org/ (perhaps thru a short path)

<Zakim> DanC_lap, you wanted to ask for uris not of running demo, but of checked-in code/docs/tests and to note that there's an HTTP header registry, so we should use it to ask the

<timbl> 1. policy served by Policy: HTTP header?

<timbl> 2. HTTP POST

<vkolovski> > HTTP header field registry

<Zakim> Danny, you wanted to ask what David means by 'proxy acts unbeknownst to the user'

vkolovski, I can't see the inteface to adding to that registry; do you? I know how to reach mnot, meanwhile.

(hmm... the postgress-TMS bit might be more relevant to the client proxy... hmm...)

<Danny> TMS in the proxy would be interesting - I'd like to see the proof generation in debug mode for demo purposes so that people can see what's happening, and even look at alternative proofs

<vkolovski> (DanC, see Registration Procedures for Message Header Fields, http://www.faqs.org/rfcs/rfc3864.html)

(I can't see the "so do this" part of that doc quickly, vkolovski )

Jim: we're distributing something that didn't used to be distributed...

(to me, it's not any more distributed in present designs than when I reviewed the proposal. I read that part of the proposal as sufficiently fuzzy as to include what Rein has turned out to be)

TimBL: you're perhaps looking at the REIN network as a network of agents, but a REIN network is a network of documents; all the execution happens in the client engine

JimH: ah... so this is similar but just a little different...

<dwood> eikeon, We need to remember to collect and remember to register all new headers used in the demo. Right now we have Controlled_by in the 401 error and probably need at least one more to pass the proof or the URL to the proof.


I think I'm interested to work on (a) proof-generating client using REIN [and twisted python] (b) openid integration


Live PAW Demo begins...

hmm... "my new policy"

DK: if you don't give an explicit policy, it's wide open [world-readable]

generic policy editor... looks like a generic forms-based RDF editor...

CWM/PML/Inference Web Integration

earlier in the morning

<DanC> PML namespace... not sure...

> Inference Web's Proof Markup Language (PML) Description

<timbl> > iw.owl in iw.stanford.edu

<dlm> > api docs

<timbl> Neither the paper nor the description have an example with the namespace

<dlm> -> ftp://ftp.ksl.stanford.edu/pub/KSL_Reports/KSL-04-01.pdf KSL-04-01 (PDF)

<dlm> > a search for proofs

<dlm> if you just pull it up you can see proofs and explainer views of the proofs

<paulo> hi Dan, yes, we met in Florida at ISWC!

paulo, timbl's deep into hacking cwm/n3 rules to turn cwm's proof notation into PML... asking questions to Deb

<li_> > my home page with my photo

<DanC> great, thanks li_

<lkagal> Hi Ding Li, how are you doing ? Hi Paulo.

<li_> Hi Lalana, I'm pretty good, nice to see you here


resuming after break...

[DebM presents. slides... pointer to be provided?]

<paulo> > my photo on my homepage

(timbl, note "Swoogle (version 2) released" -- li's homepage )

step 1 in IW registry seems to be "send a message to pp@ksl.stanford.edu" -- http://iw.stanford.edu/documents_registering.html . would be nice if step 1 was "publish a description of your system in RDF in world-readable http space)

slide Explanation Application Areas

Gerry: is that [KM from utexas.edu] Boyer-Moore?

Deb: no; Porter.
... "Task processing" e.g. 'why isn't my laptop here yet'... answer: because I'm waiting for the 3rd quote.

DanC: ooh interesting. [sounds relevant to "why isn't SPARQL a standard yet?"]

> Generalized Modus Ponens with Unification

hmm... one of the question we've been discussing is URIs for things in the IW registry

I see URI: http://inferenceweb.stanford.edu/registry/DPR/GMPwU.owl#GMPwU

can I register http://www.w3.org/2000/10/swap/log#implies ?

<RalphS> exit

TimBL: what services does the registry support?
... just giving info?

DebM: we're trying to make it easy to register stuff; our explanations get better and better as more stuff is registered.
... we like to get info about who wrote it, when...

TimBL: that should be in the ontology, right?

DebM: that seems straightforward...

<DanC> > a page on GMP

<DanC> ... that refers to http://inferenceweb.stanford.edu/registry/DPR/GMPwU.owl#GMPwU

DanC: so if I want to register log:implies, can I do so without renaming it (in URI space?)

DebM: yes.

slide [something involving trust]

DanC: do you say what you trust ?WHO about?

DebM: not yet. future/ongoing work. iTrust 2005
... ... for finding PML, finding PML with a particular conclusion

(hmm... using SPARQL for that?)

DebM: example: suggesting wine with food... demo http://iw4.standford.edu/iwbrowser/...

<dlm> > a food/wine example

<dlm> > a food/wine proof tree

<Danny> welcome, Tim

<lkagal> Hi Dr. Finin.

<finin> hi

<finin> li ding suggest I drop in

timbl's looking at http://iw4.stanford.edu/proofs/tonys/tonys/tonysns5_0.owl as pretty-printed N3

<Danny> Tim, you're welcome to call in, too, if you like.

<Danny> do you have the tel#?

<finin> no, what is the number?

+1-617-761-6200 , I think

<finin> thanks.

<Danny> yes, dial that number and then use the passcode, followed by #

<li_> ok

<li_> I'm indirectly connected from stanford

<Danny> Tim, in the room here at MIT you have all the people on the IRC channel plus Gerry Sussman and Yosi Scharf

<finin> ok, thanks for the info.

> iwregistrar

<DanC> ^ the one that's easy to find

<DanC> you could be more clear that the page isn't written by you; it's signed by "pp@ksl.stanford.edu" and doesn't seem to say who actually wrote it

<DanC> timbl, this PPDR looks like a candidate for a language for describing all these rules. http://www.ksl.stanford.edu/KSL_Abstracts/KSL-04-04.html

<DanC> found it from the registry entry for Generalized Modus Pones with Unification

<paulo> Dan, PPDR is now called InferenceMetalanguage (or InferenceML) and Inference Web does have a PML checker based on InferenceML.

<DanC> ooh... really? pointer to the code?

<paulo> > documents_checking.html

<dlm> > @@something in the registry

<Danny> > @@something else? in the gret

<dlm> > CWM, in stanford.edu

<DanC> does every pointer from PML to an inference engine have to use a stanford.edu URI?

<li_> no

<li_> those PMLs are generated at stanford

<li_> so if you handwrite a PML

<li_> you can use your own URI using namespace beside stanford's one

<scribe> ACTION: DanC to copy http://inferenceweb.stanford.edu/registry/IE/CWM.owl to w3.org/swap space

<li_> however, you need to maintain your metadescription of your rule

<timbl> > tim's attempt at PML from cwm

<DanC> maybe paulo can type in here what we should have? or is this too constraining?

<jhendler> tim - http://www.mindswap.org/2002/rdfconvert/

<timbl> > swap/test/reason

<timbl> > t4.pml

<DanC> so is there a form or something where we can put http://www.w3.org/2000/10/swap/test/reason/t4.pml and get it explained?

<timbl> iw:source

<finin> gotta go, bye

<dlm> Requested URL is unavailable for loading from the Web. is the error we are getting

<paulo> > as a template/example of PML

<DanC> looking, paulo

<timbl> Check out t4.pml now has:

<timbl> <DirectAssertion rdf:nodeID="b1">

<timbl> <source rdf:resource="t4.n3"/>

<timbl> </DirectAssertion>

dlm: need registration and cwm pml generation...

djw: I'd like dlm to follow up...

<DanC> > iwbrowser

<DanC> "This URL does not contain an IW document."

<DanC> can we have a clue added to that diagnostic?

<timbl> > to-pml.n3, the converter from cwm proof to PML

<scribe> ACTION: dlm to get cwm generating PML that can be seen in iwbrowser, with timbl

demo possibility... connect to iwbrowser...

<DanC> (cwm doesn't have a persistent cache, eikeon )

<DanC> (hmm... well, for web documents, it sorta does. but not for its inference working store)

<eikeon> That's the point.

<DanC> you have some straightfoward way to add one?

<eikeon> I think one could be added without much work, yes.

<DanC> wow... you have more confidence about playing with cwm internals than I do.

or do you see some way to do it without cutting cwm open?

<eikeon> I haven't looked at the cwm code for some time now. WIll need to look again first.

Date: 24 Aug 2005

Demo Hacking

<DanC> ok, so we've got eikeon , DaveW, Vlad, and DanC here in G8xxx.

<DanC> emacs can't find the svn command... where does it look? how do I find out?

<DanC> a goal, from JimH: demo working with rein and cwm with the pychinko engine

<DanC> proxy ideas...

<DanC> > Amit's Web Proxy Project

<DanC> DK: demo relies on twisted, rdflib, kid templating language

<DanC> pbjam:~/paw/www/2005/demo connolly$ grep url= .svn/entries

<DanC> url="svn+ssh://mindswap.org/home/svn/paw/www/2005/demo"

<eikeon> > redfoot 2.0 README

<eikeon> > redfoot dev README

<DanC> starting case... girl scout photo... if you're in the troop, you can see the photo

<DanC> I get 404 @ http://www.policyawareweb.org/2005/rein/rein

<eikeon> > Redfoot-2.0.0.dmg

<DanC> pbjam:~/paw/www/2005/rein connolly$ python2.4 ~/w3ccvs/WWW/2000/10/swap/cwm.py requestE.n3 --chatty=20 --think --filter=engine.n3 >,x

<eikeon> > session:hexdigest

<eikeon> > auth.login_process

<DanC> photo indirection TODO

<eikeon> > PAW demo README

<DanC> ^steps to get demo working, verified by DavidW

<DanC> RFE: cwm should warn when prefixes get redefined

<DanC> well, we got as far as:

<DanC> ~/paw/www/2005/rein connolly$ python2.4 ~/w3ccvs/WWW/2000/10/swap/cwm.py judyWantsPhoto.n3 troop42.n3 --chatty=50 --think --data 2>,e

<DanC> generating...

<DanC> <#_g0> reine:canAccess < http://www.policyawareweb.org/group-photo.jpg>;

<DanC> session:hexdigest "365e00b0d644076bd8a9ddef530d91c7" .

<DanC> what's the equivalent of "cvs -n update" in svn? maybe svn status?

<DanC> yes, svn status

<DanC> .

<DanC> .

<DanC> .

<dwood> Success: Redfoot installed and running from source on my Mac PB G4

demo group report

DanC walks thru judyWantsPhoto.n3 request, troop42.n3 policy, foaf auth stuff, including details of how authentication and authorization relate.

<vkolovski> TODO: exploit username (or sth) as hint so don't have to search through all foaf files for possible members

> troop42.n3 policy

<eikeon> > check.py

issue: pruning proofs

next steps: multiple troops..

LK has enough info to carry the ball on girlscout policy stuff for a few yards

<lkagal> ACTION: LK to complete Rein and fix check.py to confirm that the proof was generated with the correct policy and assumptions.

<dwood> DanC_lap: Did you choose a proxy to use?


report from cwm/pychinko group

TBL: cwm/pychinko integration issues share a lot with rdflib integration issues

cwm/pychinko/rdflib integration

tbl: ... you could use StoredStatement

> swap.formula.StoredStatement

DW: you're talking about rete as if it's a store...

TBL: yes, it is...

YS: could you build the rete differently? with, for each triple { a b c } another alpha nodes for { a ?x ?y } and { ?x b c } and { a b ?x } and so on?

TBL/Yarden: maybe, but why?

TBL: sounds like rete engine has the same interface as swap.formula [?]

while TBL and yarden [sp?] go nutso at the whiteboard...

<DanC> dwood, about strong crypto in this architecture...

<DanC> > fairly serious cwm crypto example

<DanC> > cwm crypto/trust example slides

<DanC> > sshAuth.py, my attempt at talking to sshAgent from python

trying http://www.policyawareweb.org/2005/demo/README ...

TODO: parameterize check.py to restrict premises to trusted policies

<eikeon> > redfoot README

<eikeon> > redfoot 2.0 README

<vkolovski> > check.py

<DanC> installing twisted...

<DanC> > TwistedSumo-2005-05-25.tar.bz2

- ZopeInterface (included in twisted distro)

<vkolovski> svn+ssh://mindswap.org/home/svn/paw/www/2005/rein

<DanC> victory... I've got the demo running running on pbjam, my powerbook laptop, at http://localhost:8080/

<scribe> ACTION: DanC to set up demo.policyawareweb.org subdomain

<lkagal> > Proof generator + checker example

<DanC> TEST CASE for redfoot: "abc"^^foo equal? u"abc" ? no.

<DanC> i.e. == must be transitive, yosi exhorts.


cwm/pychinko work going forward

<scribe> ACTION: TimBL to organize work on cwm (e.g. finish --why and check.py and pychinko integration of some sort) with danc, yarden, yosi, vlad, daniel.

TimBL: can we schedule cwm work time?

several: hard, since class schedules are incoming

TBL: ... weds... fri...

DanC: yeah, weds / fridays...

TimBL: 2pmET Wed 31 Aug to start?
... for 1.5hrs

YK: I have a class up to 2pm Weds...

RESOLUTION: to meet 2:30pmET Weds 31 Aug for 1.5hrs to advance cwm work. timbl, danc, yarden, vlad, daniel.

Meeting and project records

RESOLUTION: DanC may publish a record of the PAW meeting 23-24aug based on editing the IRC log with some discretion [assuming comments sent by reviewers will be addressed]

JimH: speaking of publishing... about the agreement with NSF...

<scribe> ACTION: JimH to send copy of grant to paw-team

demo work going forward

JimH: on the demo side, that's... [several..]
... demo work... lalana, daniel, david ... synchronous phone call thing?

several: yeah...

David: yeah, I think we have a schedule... running by Oct.

DavidW: for 30min 1pm ET ... [thursday?]
... for Tue 30 Aug 1pmET for <=30min?

RESOLUTION: DavidW et. al. to meet about demo work Tue 30 Aug 1pmET for <=30min


DavidW: I often use free[somthing?]

> How to reserve the W3C teleconference bridge

<dwood> PAW Demo telecon requested for Tue, 30 Aug at 1300 EST - will email when they respond

next all paw-team meetings

JimH: so the cwm/timbl and demo/david teleconference seem to obsolete the monthly telcons...

DJW: how about a teleconference checkpoint around the Oct milestone

JimH: how about 31Oct... bad for DJW...
... 27Oct...?

JimH/DJW: 28Oct...?

DJW: 28Oct... 10amET?


<scribe> ACTION: DJW to reserve bridge for all paw-team telcon

RESOLUTION: all paw-team telcon Fri 28 Oct 10amET postscript: perhaps 1.5hrs.

discussion of a next thingy in December after W3C AC meeting...

DJW reserving telcon code PAW1

5-7Dec looks hard for several...

JimH: Jan 2006?

TimBL: systinet[?] in Jan 2006 on semweb/lifesci...

vlad: classes start 25Jan 2006

DJW: 11-12 Jan?

<timbl> > conference in jan

DJW: 18-19 Jan?

DC has no conflicts any time in Jan

PROPOSED: to meet ftf 18-19 Jan in Maryland...

RESOLUTION: to have a 3rd PAW ftf meeting 18-19 Jan in Maryland; location contingent on confirmation from timbl




postscript: cwm APIs/interfaces: terms, unification/pattern matching

<DanC> { ?CITY geo:in ?COUNTRY } ??unify?? { :kc ?PROP ?X }

<DanC> unifier: ?CITY => :kc ; ?PROP => geo:in; ?X => ?COUNTRY

<DanC> { forAll ?X { ?X a OddNumber } => { ?X divisibleBy 2 } } => { something is ok }.

<DanC> { ?z a OddNumber } => { ?x divisibleBy 2 }.

<DanC> { ?z a OddNumber } => { ?z divisibleBy 2 }.

<DanC> .

<DanC> .

<DanC> { @forAll ?CITY, ?COUNTRY. { ?CITY geo:in ?COUNTRY } => { a b c} } => { g h i }.

<DanC> { kc ?PROP ?X } => { kc a funplace }.

<DanC> { kc ?PROP ?X } => { a b c }.

<DanC> { @forSome ?CITY, ?COUNTRY. { CITY geo:in COUNTRY } => { a b c} } => { g h i }.

<DanC> idea: says for combination of semantics + includes

<DanC> e.g. { <geoSparqlSvc> log:says { ?place zip "66999"; lat ?x ?long ?y } } => { ?place lat ?x; long ?y }

<DanC> phpht

<DanC> was confused with conclusion + includes

<DanC> tbl: I think that's in there. supports.

Summary of Action Items

[NEW] ACTION: DanC to copy http://inferenceweb.stanford.edu/registry/IE/CWM.owl to w3.org/swap space
[NEW] ACTION: DanC to discuss WS-Policy mapping and GRDDL with Vlad
[NEW] ACTION: DanC to set up demo.policyawareweb.org subdomain
[NEW] ACTION: DJW to reserve bridge for all paw-team telcon
[NEW] ACTION: dlm to get cwm generating PML that can be seen in iwbrowser, with timbl
[NEW] ACTION: JimH to send copy of grant to paw-team
[NEW] ACTION: LK to complete Rein and fix check.py to confirm that the proof was generated with the correct policy and assumptions.
[NEW] ACTION: TimBL to organize work on cwm (e.g. finish --why and check.py and pychinko integration of some sort) with danc, yarden, yosi, vlad, daniel.
[NEW] ACTION: YK to schedule a meeting with TimBL and Bijan to discuss n3 semantics

DanC and DavidW for JimH and DJW, 25 Aug 2005
Minutes formatted by David Booth's scribe.perl version 1.127