Policy Aware Web Project

31 Oct 2005


DanC, TimBL, David_Wood, lkagal, Vlad, DK, Danny




agenda review, scribe

<DanC> Scribe: DanC

DW: need to leave at 10:45 to get medical attention for my wrist

<timbl> +Vlad, Dan K coming

<DanC> Zakim, who is on the phone?

<Danny> 1 - update on grant status stuff - JimH

<Danny> 2 - report of cwm team (status, issues, next steps) - Tim

<Danny> 3 - report of demo team (status, issues, next steps) - DanC

<Danny> 4 - Publication plan - Jim

<Danny> Key topics: N3 Semantics, CWM behaviors, ...

<Danny> Target venues:

<Danny> - WWW 2006

<Danny> -elsewhere

<Danny> 5 - TAMI update - Danny

<Danny> 6 - Next Meetings:

<Danny> -confirming 18-19 January 2006 at UMD

<Danny> -ongoing task force meetings

<Danny> -who's going to ISWC

<scribe> Scribe: DanC

update on grant status stuff - JimH

JH: we filed our filed our annual report to NSF. We're in good standing
... we don't have a demo scheduled soon...
... official visit will probably happen early '06
... we're a little low on publications vs where we should be, but they probably won't "bash" us too badly on that. But for next year, we'll really need to do better.

report of cwm team (status, issues, next steps) - Tim

TBL: we started with the challenge of cwm --why on the girl scout case lalana built at the Aug ftf...
... as of a couple days ago, cwm --why works, as well as check.py, on that case...

<timbl> --n3=B

TBL: with a caveat about bnodes... had to [skolemize] them...
... code is fragile/raw... brand new...
... meanwhile, dan's proxy ran into trouble with lalana's rules, and there are new rules...

(a) cwm --why is working on the girl scout case

though we ran into a glitch with rein/engine.n3

when we plugged it into the proxy

(b) pychinko is using the mainline rdflib; no more branch

(c) pychinko's built-in support is coming along:

<vkolovski> btw, pychinko is able to load the troop policy file and to

build a proper rete net from it

<vkolovski> but the problem comes from the engine.n3 file, because it

uses log:conclusion and conjunction

(d) it's pretty clear now how pychinko can fit into cwm

TBL: yes, we can start on fitting pychinko into cwm

Vlad: yes, I intend to start on that. shouldn't take that long.

"We'll have a cwm/paw dev meeting 4pET Fr 4 Nov, DK at risk." -- http://www.w3.org/2000/10/swap/doc/plans.html

<DanC_> gold star timbl and vlad :)

report of demo team (status, issues, next steps)

DK: we were just hacking last night... got the proxy talking to the resource manager, with a fake proof...

$ http_proxy= wget --http-user mindswap --http-password XXX 'http://demo.policyawareweb.org/images/award.jpg'

DanC: we're sending the proof in a header on the HTTP GET, so the length of the proof is an issue

TimBL: the proof is 300 KB

<DanC_> (I've been thinking we should use a backward chainer. was gonna try euler on our use case)

DK: send it in the body of the GET?

TimBL: could the server say "put your proof here" perhaps?

DC: hmm.. maybe

DK: we've got check.py integrated into the resource manager...

DC: tim, did you say the proof takes 1.5min to check?

<timbl> 1min 3 sec

TBL: yeah... a minute or so... dunno how much is diagnostics

DC: how about using a backward-chainer... euler?

TBL: we should be able to adjust cwm so that it only has the relevant parts of the proof

JH: there's been some work on explaining fwd-chaining proofs that prunes...

TBL: cwm only spits out a proof of what it prints out... [news to me]... but the structure of the proof includes log:conclusion...

[missed some there]

DK: we should be able to put all the pieces of the girl-scout case together today: pf gen, pf check, proxy, resource manager

DC: 3 policies? we've only tested one

LK: all 3 policy files contain the same rules

JH: :-/

DW: we have a regular telcon slot... Tue 1 Nov is the next one
... next milestone is ftf mid Jan... goals for that time?
... telcon before then?

DC: I think it would be good just to address outstanding engineering issues by Jan... to get pf gen/check down to 1 or 2 seconds

JH: I have a [missed] early Dec
... in early Dec, I'm giving a cwm talk where I have to show results. workshop with NSF people etc.
... I expect maybe our PM or more likely somebody from her program

<vkolovski> btw, I will be giving a PAW talk at the policy workshop at ISWC

DJW: so... show some progress in early Dec and target realistic performance for Jan

(Mike P...i? who's that?)

JH: it's also possible that Mike P's replacement will come and ask for something. [?]
... screen dumps would work for early Dec. "it works but takes too long" is OK progress to report

DW: you want that as a deliverable of the demo group? OK.

Publication plan

Key topics: N3 Semantics, CWM behaviors, ...

VK: WWW conf submission deadline is 6 Nov

JH: there is no citeable cwm paper

DC: the tutorial isn't citeable?

JH: no, not peer-reviewed


<DanC_> (I want to do the authoring in XHTML, please. I can produce LaTeX from it. http://www.w3.org/2004/04/xhlt91/ )

TBL: I asked Gerry for help on formal semantics... he gave a somewhat dismissive answer...

(dismissive of formal semantics, not of timbl)

<timbl> suggested taht an operational semantics was all tht made sense ,,, more or less

(vkolovski, "academic style" often means TeX. Let's do the collaboration in XHTML, OK?)

JH: another target is the web semantics journal. hmm... turn-around time might be an issue

<timbl> Important in cwm etc to get the *phioliophy* of what we are trying to get out ther up front as to why we aren't chasing the same traeoffs as previous systems.

<timbl> like /DesignIssues/Diff



JH: so that's cwm papers...

<DanC_> (what of "KR" conference?)

JH: another writing assignment DJW and I are ware of is TAMI...

<vkolovski> KR would be a nice venue for our n3 semantics paper,

JH: so that's the topics... (1) cwm, [missed 2 and 3?]

<vkolovski> however the deadline is early nov., too

JH: so that's the topics... (1) cwm as problem solver (2) approach to policy awareness (3) formalization of N3
... we have a book chapter of (2) but need to revise that

DJW: how about a presentation at Jan PAW ftf of cwm [stuff]?

<Danny> N3 Logic design philosophy/motivation

TBL: umm... ok, but it's important to put [it] in the paper... the Diff paper reviewers didn't get some of the main points... our system is crude, but looking stuff up in the web is novel... [?]

JH: "cwm: a problem-solver for the Web" or something like that

<Danny> Paper -- CWM: A Problem Solver for the Web

<Danny> :-)

<timbl> s/crud/crude/g :-)

JH: my 1st impression of cwm was "so these built-ins are novel, but so what?" but then I learned there's more to it

TBL: I'd like some brainstorming... if I could spend 45 min on the phone after JH reads /N3Logic ...

<timbl> The dsign philosophy of gatheringrules from all over the place and getting the result you want

[we get a preview of such brainstorming...]

DJW: how about ACTION JH: report back on brainstorming with TBL on N3Logic

<scribe> ACTION: JH to report back on brainstorming with TBL on N3Logic

VK: about our approach to policy awareness... I'm presenting at the policy workshop at ISWC...
... I expect peertrust folks will be there...
... anybody else will be there?

LK: I'll be there. [chairing]

DJW: hope to attend part. Monday.

TBL: can't be there. I'll be in the industry part...

<timbl> I will be in the Industry Day .. i may try to get out but it is unlikely ... launching new working groups etc

LK: I think our paper is late in the day 4:xx...

DJW: that's hard

next meetings

DJW: to re-iterate: 18-19 Jan at UMD

JH: invite DebM to Jan/UMD meeting? think about it. soon.

TAMI update

LK: TAMI will be a tool to record audit trail proofs

(fyi, the TAMI group's weekly rhythm is a sort of class. http://groups.csail.mit.edu/dig/2005/09/6.898/ )

<timbl> Do we have a 'plan' page in which I can find all these pointers etc

LK: we've been working on a "deadbeat dad" law case... what data is allowed to be used where



<scribe> ACTION: DanC to add link to lk's REIN examples from policyawareweb.org

TBL: I'd like a page that collects [missed]

<timbl> all the test data whcih cwm is expected to handle, so I can retest it

<lkagal> I'll be updating this page with more TAMI examples : http://groups.csail.mit.edu/dig/2005/10/TAMI/examples/

expected by the PAW project?

Postscript: Writing assignments

<scribe> ScribeNick: nobody

<lkagal> DanC, TimBL, Vlad, I'd like to help with the cwm paper. Would that be OK ?

<DanC> yes, I'm counting on it

<lkagal> :)

<DanC> have you started reviewing http://www.w3.org/DesignIssues/N3Logic ?

<DanC> I think " The log: namespace has functions " assumes the reader shares a rather parochial understanding of "namespace

<DanC> "

<DanC> hmm... tim, what would you think about using the paw svn space to edit N3Logic? we'd all have write access then

<lkagal> Yes. However, I believe it was decided to split up the paper into 2, (i) cwm functionality, and (ii) N3 semantics/logic. I believe we're starting with the cwm functionality paper which will be made up of the tutorial and the rule wkshp paper, http://www.w3.org/2004/12/rules-ws/paper/94/

<DanC> I don't think N3Logic has anything about cwm functionality, does it?

<timbl> I don't see what is interesting about CWM which isn't very much to do with N3 logic

<DanC> tim, I think "cwm, a problem solver" is a totally diffent paper from "formal semantics of N3"

<lkagal> JH: so that's the topics... (1) cwm as problem solver (2) approach to policy awareness (3) formalization of N3

<timbl> Feels like teh n3 logic paper is background you need for cwm, and cwm the the mativational paper you need to understand wht n3 logic.

<lkagal> I think the idea is to first describe what cwm does before going into why.

<DanC> tim, there are people that don't need any motivation to pore over a paper on N3 logic. Look at Herman ter Horst's review of SPARQL. http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2005Oct/0068.html

<DanC> in sum, "get all that tutorial stuff out of the way so I can study the math"

<timbl> Yes.

<DanC> so yes, some people will use one paper as motivation to read the other. but they have different primary audiences, I think.

<lkagal> So, I still think the N3 formal semantics paper would be perfect for AAAI's track on AI & the Web. The cwm functionality paper can be sent to ISWC or a WWW workshop - I don't think it would make it to a journal.

<timbl> Ok.

<timbl> So who is going to write that paper -- be chief cook?

<DanC> timbl, you're chief cook until you delegate

<lkagal> I'd be happy to help put the paper together.

<timbl> Well, I though I heard Vlad say that he was going to write it, and may have heard Lalana suggest she would, and I would be happy to delegate to either or both

<vkolovski> just to jump in, heard you talking about it

<vkolovski> I think it's a good idea to have a cwm systems paper for the journal of web semantics

<vkolovski> like jim suggested

<timbl> Ok, so is that a plit then - Vad do the cwm paper ad Lalana do the N3logic paper?

<vkolovski> i can take the initiative for that paper, but will need help

<timbl> Yes, certianly.

<DanC> +1 "Vad do the cwm paper ad Lalana do the N3logic paper"

<timbl> I am happy to give help where I can

<DanC> and do it in XHTML pretty please (though I'll understand if you say no)

<vkolovski> Danc, I will have problems with all of the equations when writing it in html

<DanC> are you sure? you can just write <samp>go nuts with latex math mode here</samp>

<DanC> how many equasions will you need in a systems paper on cwm?

<DanC> equations

<vkolovski> oh i see - what about the journal proceedings templates

<timbl> Ok, is that agreed, then? Vlad, Lalana? Vlad to be chief writer on the cwm systems paper and Lalana on the N3 logic one?

<vkolovski> timbl, it's fine with me

<timbl> Thanks.

<DanC> I'm pretty sure I can adapt the tools to other latex styles, vkolovski

<DanC> it's probably just a one line change

<lkagal> Yes, its fine with me too. Vlad, if you need some help with the cwm paper, let me know.

<timbl> ACTION: Vlad chief writer on the cwm systems paper for the journal of web semantics.

<vkolovski> Danc, I will investigate the option.

<DanC> vlad, what's the deadline for that? I guess journals don't have deadlines?

<timbl> ACTIO: Lalana be chief writer on an N3 logic aimed at AAAI's track on AI & the Web.

<DanC> if you can drop a pointer to journal of web semantics "instructions for authors", pls do

<vkolovski> It's a rolling submission - it takes more time to get the results back though

<vkolovski> Jim mentioned he will talk to the JWS people at ISWC

<vkolovski> ok will do

<vkolovski> http://www.websemanticsjournal.org/ps/pub/welcome

<vkolovski> http://authors.elsevier.com/JournalDetail.html?PubID=671322&Precis=DESC

<DanC> "the author(s) will be asked to transfer copyright of the article to the Publisher" BZZZT!

<vkolovski> it's always like that with conference journal publications

<DanC> I don't see a latex style; they ask for PDF, with certain layout and references styles. Do they re-key the paper?!?

<DanC> it _was_ always like that with conference journal pubs.

<lkagal> Most confs and journals do provide a latex style.

<DanC> "My artwork is on a 3.5" disk, CD-ROM, ZIP disk or JAZ disk." chuckle.

<vkolovski> :)

<vkolovski> what do you mean, i just signed a copyright release form for the ISWC paper

<vkolovski> what is the big problem with latex?

<vkolovski> it's easy to edit, no?

<DanC> no

<DanC> the only way to read a latex paper is to print it. ugh.

<DanC> latex syntax is incredibly arcane. look at latex2html.pl

<timbl> The Southamton people have EPrints and have been battling the publishers hard.

<DanC> "Tables must be typed on separate sheets and should have a short title and an arabic number." ?!? this is for a journal with online delivery? hello, Elsevier, welome to the 21st century.

<vkolovski> "Quarterly electronic publication; yearly archival print volume in December"

<vkolovski> at http://www.elsevier.com/wps/find/journaldescription.cws_home/671322/description#description

<vkolovski> also, this:

<vkolovski> Open-source software and tools that help to advance the field more rapidly, will be published and demonstrated based on a rigorous review and selection process

<vkolovski> considering this, we're in good shape :)

<DanC> yes

<timbl> Ooops.. seems I wrote the code to check the valdiity of a rule's evidence twice ... which would increase the time significantly as it is recursive. Now 25s

Summary of Action Items

[NEW] ACTION: DanC to add link to lk's REIN examples from policyawareweb.org
[NEW] ACTION: JH to report back on brainstorming with TBL on N3Logic
[NEW] ACTION: Vlad chief writer on the cwm systems paper for the journal of web semantics.
[End of minutes]

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