13:25:51 <RRSAgent_> RRSAgent_ (~DAML-rule@tux.w3.org) has joined #DAML-rules
13:25:51 <pfunk.w3.org> Users on #daml-rules: RRSAgent_ em @tim-daml 
13:26:19 <tim-daml> - the agent has already loaded all knowledge about the site's info.
13:26:36 <tim-daml> RRSAgent, Richard Fikes is talking abou tneed for reasoning on the Sweb.
13:26:51 <tim-daml> Needed for
13:27:07 <tim-daml> - ontology builders (inconsidtency detection, classification)
13:27:15 <tim-daml> - web sit ebuildefrs 
13:27:21 <tim-daml> - Users seeking info
13:27:31 <tim-daml> - Users seeking services to take answers
13:27:46 <RalphS> RalphS (~swick@tux.w3.org) has joined #daml-rules
13:28:05 <tim-daml> Simple examples of need for reasoning:
13:28:14 <RalphS> RalphS has changed the topic to: RRSAgent public log is http://www.w3.org/2001/07/20-DAML-rules-irc
13:28:16 <tim-daml> - ask about parent, system knows child.
13:28:20 <RalphS> RalphS has left channel
13:28:51 <tim-daml> - ask about parent, system knows father  (subclass)
13:29:28 <tim-daml> Query examples
13:29:56 <tim-daml>   - example with forwarrd chaining rules for subclass
13:30:14 <tim-daml>  - examples using a restriction
13:32:11 <tim-daml>   In these examples, the DAML ontology isused, plus the DAML axioms.
13:37:15 <tim-daml> (Someone is on the bridge)
13:39:39 <tim-daml> Richard gave examples of four or so projects which use reasoning in the DAML project.
13:39:45 <tim-daml>  (didn't include cwm)
13:39:55 <tim-daml>   FaCT, etc
13:44:29 <tim-daml> Q: You are assuming that users wiull make very precise queries (who is the chief justice of the US Supreme Court)
13:47:37 <tim-daml> ------------------------------------
13:47:47 <tim-daml> DAML+OIL Overview, Ian Horrocks
13:51:35 <tim-daml> - Operators restricted so decidable - consistent ith layers architecture between RDF and undecisable layers
13:51:50 <tim-daml> - Therefore, allows reasoning services can be provided
13:52:08 <tim-daml> Why reasoning?
13:52:19 <tim-daml> - Ontology design, checking consistency
13:53:15 <tim-daml> - ontology integeration
13:53:43 <tim-daml> - Ontology deployment
13:56:04 <tim-daml> Tim Q:  Note that resaoners are useful for actul data processingtoo!
13:56:15 <tim-daml> Examples of Reasoning
13:56:58 <tim-daml>  - [ finding inconsistency in data, something both "high' and "low" ]
13:57:17 <tim-daml>    That was an exapled using OilEd
13:57:38 <tim-daml> Examples of things you can't do
13:58:05 <tim-daml>   - limitatilons of the logic having a tree model property, [???]
13:58:18 <tim-daml>  - no property constructores
13:58:25 <tim-daml> - no variables
13:58:32 <tim-daml>  - Only unary and binary predicates
13:59:09 <DanCon> DanCon (~connolly@tux.w3.org) has joined #daml-rules
14:15:05 <tim-daml> [...]
14:15:40 <tim-daml> steve/cyc man - suggests yes, use   template => tempalte with varabes as fiskes suggested, but also
14:16:08 <tim-daml> plese add builtins for string and arithmetic, group membership, pattern matching etc.
14:16:58 <tim-daml> Pat: Designer reasoner first then design langauge for it.
14:17:23 <tim-daml> Pat: You can do a queryt in daml as Ian H outlines - you can design a class of objects matching query using restriction.
14:17:40 <tim-daml>    however this is a weird way of encoding it.
14:18:32 <tim-daml>    You would want an automatic tool.
14:18:44 <tim-daml>   The same expressing power can be done in a number different way.
14:18:57 <tim-daml>   Logic people find calsses and cardinality difficult at first.
14:19:30 <tim-daml> Drew: SQL is an obvious choice -- so many people use it. You won't get anywhere is you insist on DAML + OIL.
14:19:33 <tim-daml> --
14:19:36 <tim-daml> Fikes:
14:19:51 <tim-daml>   Pop up, we should for a working group - what shoul it do?
14:20:05 <tim-daml>   We can have us a nd a website and mailing list and stuff.
14:20:11 <tim-daml>   What else shoul dthat group do?
14:20:29 <tim-daml>   One thing is a query language, a second is a language for justifications.
14:20:44 <tim-daml>   [Proofs]
14:20:54 <tim-daml>   Ihave asked Ben to seed the discussion on this.
14:21:24 <tim-daml>    From my point of view on hte SWeb we will be asking queries of agents we don't know
14:21:44 <tim-daml>   so they must be able to return more than just the answer (17.5) we will need
14:21:52 <tim-daml>    a justification of how they arrived at it.
14:21:55 <tim-daml> ------------------------------------------------------
14:21:56 <tim-daml> Benjamin Grosof/Sloan
14:22:04 <tim-daml>        Justifications in Hybrid Reasoning
14:22:05 <tim-daml> .
14:22:34 <tim-daml> We want to support multi[ple KRs, including FOL, DLs, etc
14:22:57 <tim-daml> Wen we webize, we will be explouting many KBs and transmitting information between them.
14:24:23 <tim-daml>    (same slide as hi spresentation yesterday -- what's hot)
14:24:33 <tim-daml>   Translation between rule systems.
14:24:46 <tim-daml>   Need inter-site communication of premises and conclusions.
14:24:59 <tim-daml>   Many systems are non-monotonic
14:25:07 <las> las (~las@p206.usnyc8.stsn.com) has joined #daml-rules
14:25:37 <tim-daml>    There is a certian amount of overturnability as to whether you trust [a non-mon reasononer's] word.
14:26:42 <tim-daml>   We should be prepared to deal with hte fact htat a lot of the information, 
14:26:57 <tim-daml> the poilicies we are able to import and trust will have nonmon aspects.
14:27:12 <tim-daml> +peter patel-schneider on the bridge.
14:28:42 <tim-daml>   Ben G/ctd: We shoudl suppoer both backward and forward reaosning.
14:29:19 <tim-daml>    we shouldn't worry aboyut the algorithm used, only the  justification, the proof.
14:29:44 <tim-daml>   Shoudl include: the source, the KR being used, (nonmon or mon).
14:31:04 <tim-daml>   We might want to burrow down into sub justifications.
14:31:21 <tim-daml>   The query may have to be rerun hte query with justification son.
14:31:32 <tim-daml>   (see his slides)
14:34:24 <tim-daml> Drew: You are making the problem too complicated.  You are argue that to really know htat stanford is having a meeting in 2002, you will ahve to do deeper reasoning.
14:34:58 <tim-daml>   I counterpropose we distinguish between reasoning I do and that other agents do. 
14:35:18 <tim-daml>   I have a trusted system has leaves I trust and define.
14:37:32 <tim-daml>  Can't I just hand people what I have ..
14:37:35 <tim-daml>  [...]
14:39:46 <las> timbl: there need to be variable levels/kinds of trust...
14:40:14 <las> ...credit card (needs to be a bill) vs. random factoids (may be ok if they're hearsay)
14:41:07 <las> Pat Hayes:  I want more info about where the information comes from....if I hear something isn't on the web from a mediocre search engine, I want to know "I didn't find it", not "it isn't there.  I want to know what sort of reasoning comes to negative conclusions.
14:42:37 <las> timbl:  two justifications:  (1) we like to be able to write a justification for, e.g., access control (in terms of membership, authorization by member rep...)  where you need to justify (show why)....fullblown search is too expensive, but you can take some hints -- look at my Stanford affiliation -- that help you heuristically find the inference path.
14:43:54 <las> Jim Hendler:  I agree, but would go further.  I may need to tell you something about where you need to go to get info....when you say you're a member of IBM, you need to use the IBM-authorized site.  If I can name the website from which you need to get some info, you can't just assert it; I need to be able to go check it.  Anchoring (like this) is extremely powerful.
14:44:17 <JimH> JimH (~newbie@p69.usnyc7.stsn.com) has joined #daml-rules
14:44:42 <las> Ben Grosof:  Tim's example is an example of the class of authorization policy; there's a lot of work in this area, including thinking about this kind of delegation.  Also people are suspicious about that app, and they need to investigaqte down to the last justification.
14:45:16 <las> Drew:  No, actually you don't trust the person, but you do trust the company and you have to trust IBM about its assertion that the person is a member.
14:45:18 <DanCon> I like the focus on justifications (aka proofs) as the pivot point of interoperability. Do all the non-mon stuff you want, as long as you're prepared to justify it.
14:45:36 <las> Tim:  There must be a statement from the organization digitally signed and asserting the membership.
14:45:51 <DanCon> after all, TimBL, your design for integrating DSig into the SWeb is, technically, non-mon.
14:45:54 <las> Drew:  You trust the members (e.g., IBM) pretty much unequivocally.
14:46:14 <las> Ben:  Typically people won't probe thse things arbitrarily far.
14:46:21 <las> Tim:  Typically one loop, then signed.
14:46:44 <tim-daml> (My second poit was that fundamentally the moment hte lnguage is undecisable, resonsing engines will not be able to be repeat the reasoning of another, so prro are the only thing can exchange) 
14:47:58 <las> Jim:  There's a combinationo f a technical and a social point here....I want to convince you that I'm the Chair of CS at MIT....I create a website, put MIT on the page 3000 times (so Google finds this)....you say I only believe it if MIT says so....I can fake this....but you say "my definition of MIT says is from the Amer Assoc Univ def of "MIT's website""....you get these groundings and notions of ownership pretty straightforwardly on the web.
14:48:06 <las> Ben:  Certificate authorities
14:48:24 <las> Tim:  Just "on the web"
14:48:54 <las> 10:37 AM: <JimH> JimH is now in the logic/ruls group - Benjamin Grosof has given a long talk about justifications 
14:48:55 <las> 10:37 AM: <JimH> Drew asks some good questions about whether they're focused on the right place
14:48:55 <las> 10:37 AM: <JimH> ... trust and justification (as a formal issue) vs. as a "social process" (my words, not his)
14:48:55 <las> 10:39 AM: <JimH> non-mon entering the discussion -- on the web it is non-mon, no matter what you try (in a certain sense)
14:48:57 <las> 10:42 AM: <RalphS> Jim, Tim is note-taking the rules discussion in #daml-rules; see its topic for the log
14:57:42 <tim-daml> --------------------------------------- break -----------------------
14:58:17 <tim-daml> Sun's Juxtaposition  jxta.org  peer-peer  is on collabra (?) -- Steve/cyc looking at it 
14:58:29 <tim-daml>    discussoing queryt delegation
14:58:40 <tim-daml>   using unary properties.
14:59:39 <tim-daml>   extensional collectional - don't backchin - definitive list - exhasutice list 
14:59:48 <tim-daml>   class member list definitive.
15:04:42 <tim-daml> ------------------------ end break -------------------
15:04:50 <tim-daml> Mike Dead on DAML Rules
15:04:53 <tim-daml> - see his slides
15:05:02 <tim-daml>   Several strawman proposals.
15:05:50 <tim-daml>   his slide has pointers to them.
15:06:05 <tim-daml> .
15:06:25 <tim-daml> Stefan Decker,   Rules Requirements --------------
15:06:36 <tim-daml> Let's classify what people can do with rules.
15:06:45 <tim-daml> See his slides
15:07:05 <tim-daml>   they include Rewriting/Transformation,
15:07:07 <tim-daml>  Matching
15:07:11 <tim-daml>  - equivalence
15:07:20 <tim-daml> - computation rules wiht arithmetic builtins
15:07:33 <tim-daml> - clss rules (changing hte class striucture)
15:07:54 <tim-daml>   (computational includes procedural attachments)
15:08:05 <tim-daml>  - Traust rules - data fro a cerain source
15:08:07 <tim-daml> - proofs
15:08:11 <tim-daml> --------
15:08:27 <tim-daml> Worry - a new rule language for evry purpose.
15:09:13 <tim-daml> Unifying ground is the rDF data model.
15:12:52 <tim-daml> His example in an N3-like but not equal sysntax.
15:13:04 <tim-daml> Q: Are queries just rules
15:13:09 <tim-daml>  - time
15:13:33 <tim-daml> A: Steve cyc:   YEs, querye sdo agregation such ans min/max, and ordering of results **
15:13:38 <tim-daml> .
15:13:41 <tim-daml> Trust rules
15:14:05 <tim-daml> RDF i snot enough
15:14:15 <tim-daml> - is a statement in this SET OF TRIPLES?
15:15:50 <tim-daml> Relationshipo between DAML+OIl and Rules?
15:15:59 <tim-daml> Bultins? Procedural attanchments?
15:16:21 <tim-daml> [tim notes also we needn remote webs ervice delagtion fo subquies]
15:17:29 <tim-daml> Steve: For procedural attachments, do you mean in otehr languages than prolog source code? I suggest any function anywhere.
15:20:30 <tim-daml> ---------------------------------------------------------------------------
15:20:37 <tim-daml> Ben G:  What'snew with RuleML
15:20:42 <tim-daml>  [see slides]
15:22:23 <tim-daml> RuleML is unonorders (This makes use of a DTD difficult)
15:26:56 <tim-daml> [missed a bit]
15:27:10 <tim-daml> Ben gives examples of ruleml
15:28:01 <tim-daml> Steve: Is reuleml expressive enough for prolog program?
15:28:53 <tim-daml> Ben: Pure polog, without builtins, and ";" which is disjunctive construct whcihsome have, then ruleml can express basic momotonic prolog.
15:29:04 <tim-daml> Steve: Is there any xml serialization of prolog?
15:29:29 <tim-daml> Ben:      dunno - maybe .  see IBM business markup language.
15:29:46 <tim-daml> [this stuff is all xml, not daml]
15:30:24 <tim-daml>   Ben: Next step is to figure out how to combine  riules wiht DAML.
15:30:40 <JimH> JimH notes off line that one of Ben's problems is that DAML (which has Berners-Lee and Hendler names on it) is getting attention while RuleML (which has his name) isn't - sigh
15:31:25 <tim-daml>   Classes and properties become unary and binary preicates, but some things in DMALcan't be epxressed in LP.
15:31:29 <tim-daml> (?)
15:31:56 <tim-daml>  Eg: There existss a filler of a property for evry instance of a class
15:38:50 <tim-daml> IanH: DMAL +Oil is very well understood, and rules are different. Thery are different subsets of first prder.
15:39:32 <tim-daml>   teh result is that then full first order.
15:39:42 <tim-daml> Pat: Some of us inc cyc are not worried baout that.
15:40:01 <tim-daml> Steve: We could have feature sets, clients and servers negotiate.
15:40:31 <tim-daml>   I would be in favor of first order - ew are first order.
15:40:38 <tim-daml>   We are slow , so I sympathize with Ian.
15:41:57 <tim-daml> Ian: We tlked about that.
15:42:08 <tim-daml>   You could have hte new langauge as the union.
15:45:03 <tim-daml> ** OPen question - whether to specify the subset of DAML+OIL can be l implemented using the language.
15:51:57 <tim-daml> N3 idea:  <-  operators, if tey take precedence over . could maybe maye at least one set of {}  in a rule unnecessary.
15:53:52 <tim-daml> em?
15:55:20 <tim-daml> JimH and Ralph and I deicded that we would create www-rdf-logic@w3.org as a gubgroup if ww-rdf-interest, just like logic.
15:55:53 <tim-daml> We need to set that up and annoiunce it to the logic list as it happens.
15:56:14 <JimH> timd-daml - assume you mean www-rdf-rules
15:57:56 <em> em here
15:58:13 <em> tim-daml, whats up?
15:59:31 <em> note: www-rdf-logic scope "W3C provides the www-rdf-logic forum as a home for detailed technical
15:59:31 <em> discussion of all approaches to the use of classical logic on the Web for 
15:59:31 <em> the representation of data such as inference rules, ontologies, and
15:59:31 <em> complex schemata."
15:59:38 <em> - http://lists.w3.org/Archives/Public/www-rdf-logic/
16:04:30 <em> not apposed to www-rdf-rules but juest wanted to note www-rdf-logic can serve discusses on rules as well... my guess is a good part of the current logic discussion will move to w3c-webont-wg (name?) when this is chartered
16:18:37 <tim-daml> [missed a lot]
16:19:43 <tim-daml> Pat: Everyone knwos that a query language for dmal+oil is daml+poil with variables
16:26:39 <las> las has quit (turning into a pumpkin....)
16:26:44 <tim-daml> Ruels and Qurery - Fikes is for query,   being different
16:27:12 <tim-daml> Fikes - 6 months to a rule language.
17:21:41 <tim-daml> (Richard Fikes ends up maintining that query is sifferent, but as a minority)
17:22:26 <tim-daml> Jim Hendler suggests that we shouldcollaborate wiht Mike Dean on cool management tools for working groups. Maybe Eric and Mike Dean should chat?
17:26:24 <em> good idea JimH
18:15:36 <danbri> danbri (~danbri@tux.w3.org) has joined #daml-rules
18:18:38 <danbri> Er, rumour has it you guys have just asked for a www-rdf-rules mailing list as an RDF Interest Group sub group...
18:19:32 <danbri> You're welcome to cc: me next time. Meanwhile, I'd be interested to hear why this doesn't fit in the www-rdf-logic charter, http://lists.w3.org/Archives/Public/www-rdf-logic/
18:20:09 <danbri> I'm not against a new list in principle, but we'll need to be clear how the roles carve up. Folk are allready too tempted to x-post rdf-logic and rdf-interest...
18:24:37 <DanCon> ugh... a separate rules list?
18:25:29 <DanCon> who's the staff contact?
18:25:30 <danbri> apparently.
18:26:08 <danbri> I wrote the rdf-logic charter pretty carefully so as to include rules in its scope; I've not idea wtf is going on.
18:26:12 <DanCon> Maintainer_Email: swick@w3.org
18:27:12 <DanCon> I could support starting a rules list only if we close the logic list.
18:30:24 <em> has www-rdf-rules been announced at the daml meeting?
18:30:50 <danbri> sorry guys, you just can't go creating RDF Interest Group sub-groups on random f2f-inspired whim.
18:31:04 <danbri> Let's put the mailing list on hold and talk about this at SW CG next week. 
18:31:16 <em> agreed
18:31:37 <danbri> I consider rdf-rules bogus and harmful until someone rewrites www-rdf-logic charter to clarify respective roles of the two lists.
18:41:52 <danbri> For the record, I think it is a smart move to break Rules out from under WebOnt, and to discuss things in a non-WG context for a bit before going standards-track.
18:45:05 <danbri> (just mailed benj. grosof, to see if he's around next week for a chat about how best to do this)
18:45:23 <danbri> if anyone's f2f with him, could you mention this?
18:54:03 <tim-daml> Danbri, JimH Ralph and Ben G and I agreed that we would start the public list.  However,
18:54:42 <tim-daml> a DAML-rules public list has also been started so it will compete for traffic.
18:54:59 <tim-daml> There is in practice a question as to wheytherthis becomes a DAML committee more than an open interest group.
18:56:07 <tim-daml> bogus until charter written, yes.
19:01:02 <danbri> It's a balancing act; if these lists are getting created, makes sense to have them under W3C umbrella. But we're drifting into territory of inventing new "w3c-process-Lite", shadowing W3C WG / IG structure with mailing lists that are "working group-ish"
19:02:01 <danbri> they could just be mailing lists, a la various e-groups discussion fora. Or they could be groups governed under some non-W3C process (eg. DAML language ctte, RSS). 
19:03:06 <danbri> What I object to is pre-announcing a forum that has this grey status: partly "just a mailing list" and partly a WG-like entity. It'll create confusion.
19:09:15 <JimH> JimH has quit (Broken pipe)
19:24:21 <danbri> danbri has quit (EOF From client)
19:30:24 <tim-daml> tim-daml has quit (EOF From client)