Semantic Web Interest Group IRC Chat Logs for 2002-10-30

00:48:14 <AaronSw> google sez " rdf is a convenient method to represent ifla frbr information"

00:51:26 <GabeW> hey, is it fair to say that OWL is just the next version of DAML+OIL?

00:51:43 <AaronSw> yeah

00:51:56 <GabeW> kewl

00:52:10 <sbp> Gabe: no, I don't work on HumanML

00:52:17 <AaronSw> ...anymore

00:52:23 <GabeW> heh

00:52:26 <GabeW> delayed response

00:52:29 <GabeW> ok

00:52:31 <GabeW> just curious

00:52:50 <sbp> .googlecount sbp HumanML

00:52:51 <datum> sbp HumanML: 91

00:53:30 <GabeW> saw your name on some docs

00:53:47 <AaronSw> sbp used the be like totally into that humanml stuff

00:53:53 <AaronSw> then he discovered he wasn't human :(

00:53:57 <GabeW> like totallY?

00:54:03 <sbp> they don't seem prone to updating information; I was listed as a TC member for months after I left

00:54:14 <sbp> Aaron is a like totally monger

00:54:21 <sbp> (here googlism, googlism)

00:54:34 <GabeW> heh

00:54:35 <GabeW> ok

00:54:52 <GabeW> just trying to get the inside scoop on what humanml is

00:55:23 * sbp never really found out

00:55:44 <sbp> like XML, it looked like a good idea at first...

00:55:57 <GabeW> heh

00:57:09 <GabeW> like several of my past jobs

03:05:46 <ronwalf> Hm, doubt dajobe would be awake this time of night...

09:53:02 <JibberJim> dajobe?

10:25:16 <JibberJim> dajobe, I get libtool: link: warning: undefined symbols not allowed in i686-pc-cygwin shared libraries compiling raptor under cygwin, known issue?

10:26:30 <dajobe> nope

10:26:39 <dajobe> hmm

10:26:44 <dajobe> try building a static version

10:26:48 <dajobe> ./configure --disable-dynamic

10:27:08 <dajobe> danbri: you've done this too, any suggestions?

10:27:59 <JibberJim> rdfdump works, after the compile, it's only a warning.

10:28:08 <dajobe> oh

10:28:16 <dajobe> hmm

10:28:23 <dajobe> take it to google? ;)

10:28:48 <JibberJim> requiring cygz and cygxml dll's aswell as cygwin1.dll over the static version (which I compiled last time)

10:29:10 <dajobe> what's cygz?

10:29:18 <dajobe> (who invents these names!)

10:29:47 <JibberJim> don't ask me, it just needed it...

10:30:48 <JibberJim> seems to have compression funcs in it from the symbols exported.

10:30:58 <dajobe> oh, gz lib

10:31:04 <dajobe> must be libxml that needs it

10:31:33 <dajobe> i see now; cygz=cyg + libz, cygxml= cyg + libxml

10:32:00 <dajobe> xml2-config --libs

10:32:00 <dajobe> -L/usr/lib -lxml2 -lz -lm

10:32:03 <dajobe> yes

10:33:55 <JibberJim> do want me to look further about where the warnings coming up, or not worry as it works ok?

10:34:13 <dajobe> it's probably some win32/cygwin linking thing

10:34:21 <dajobe> you can email me the output if you like

10:35:55 <JibberJim> okay, I'll re-make it when I next leave the machine...

12:36:21 <ronwalf> dajobe: In raptor, where are the global references to world defined?

13:14:41 <aevirex> hi guys

13:16:00 <aevirex> hello???

13:16:24 <JibberJim> hi

13:16:37 <aevirex> hello...whats up?

13:19:25 <aevirex> hello?

13:24:08 <aevirex> is anyone therre????

13:24:42 <dajobe> ronwalf: eh?

13:28:02 <ronwalf> dajobe: Raptor is failing to build on me

13:28:14 <ronwalf> it's throwing out all sorts of 'undefined' errors

13:28:58 <dajobe> email me the logs

13:29:07 <ronwalf> config.log?

13:29:12 <ronwalf> Or the output?

13:29:20 <dajobe> well whatever is "throwing out all sorts of 'undefined' errors"

13:29:39 <ronwalf> Ok, let me start from scratch

13:29:57 <dajobe> I could believe raptor inside redland might be broke

13:30:03 <dajobe> since I'm changing the interface between them

13:30:09 <dajobe> raptor standalone works fine

13:30:51 <ronwalf> Yes, it's the INSIDE_REDLAND stuff

13:30:55 <ronwalf> I checked that last night

13:31:00 <aevirex> can ya say me what exactly is the topic in this channel???

13:31:07 <dajobe> ronwalf: all that stuff will go

13:31:16 <ronwalf> ah

13:31:30 <dajobe> aevirex: rdf and related techs, we tend to wander into other web technology too

13:32:39 <ronwalf> Mail is off

13:32:56 <ronwalf> Oh, huh, and that wasn't on one of my home machines

13:33:38 <ronwalf> ah well, same error

13:33:54 <aevirex> cooooool...very interresting...really

13:34:21 <dajobe> ronwalf: yeah, log looks like it is broke in that area. I'm not surprised. If I had a little time this week, I'd fix it.

13:34:27 <aevirex> but(sorr yim not an american)what means "RDF"

13:34:40 <dajobe> don't be sorry, neither am i :)

13:34:42 <aevirex> and i cant speak english very well

13:34:55 <dajobe> try www.w3.org/RDF/ to start

13:36:22 <aevirex> vocabulary description language????

13:37:03 <ronwalf> Kindof

13:37:52 <ronwalf> Sometimes I feel I need a docterate to explain what RDF is

13:38:41 <aevirex> sorry...there was an article "RDF vocabulary discription language"

13:38:49 <ronwalf> From the most basic perspective, it's just another data exchange format, built around making it easy for programs to communicate

13:39:28 <ronwalf> But I have a feeling that if I say "Data Exchange Format" around here, I will be shot

13:40:51 <ronwalf> There have been lots of hard work done to make RDF useful for things like Artificial Intelligence

13:40:51 <ronwalf> Well, kinda

13:40:51 <ronwalf> I'm getting things wrong

13:42:41 <aevirex> cool...well i searched with google(microprocessor irc) and i found this channel!

13:48:18 <aevirex> hello?

13:50:46 <ronwalf> Sorry, there are a lot of people here, but we tend to just leave the window open and do other things

13:51:10 <aevirex> ohh...i do so...

14:01:45 * aevirex is away: (Auto-Away after 10 mins) [BX-MsgLog On]

14:14:02 <aevirex> bye...well i try menuetOS

18:28:53 * tim-away wonders whether jeff heflin ever hangs out here under some name

18:57:14 <dajobe> my rdf core wg mail just broke 10,000 messages and broke my mail program

18:57:19 <dajobe> maybe it is for the best...

19:02:38 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

19:07:16 <attila> hi sandro

19:09:58 <sandro> hello.

19:12:20 <sandro> File "/home/sandro/cvs/cvs.w3c.org/WWW/2000/10/swap/llyn.py", line 2593, in tryHeavy

19:12:20 <sandro> raise BuiltInFailed(sys.exc_info(), self ),None

19:12:20 <sandro> llyn.BuiltInFailed: Error during built-in operation

19:12:21 <sandro> 20) short=0, 0_work :: "Any stat" semantics _g11(_g11)?.

19:12:22 <sandro> because:

19:12:25 <sandro> " Unable to access document <md5:16117d0b44fad96c609940888bd1aa87>, because:\n [Errno url error] unknown url type: 'md5'"

19:12:30 <sandro> sorry

19:14:55 Disconnected from irc.openprojects.net (ERROR :Closing Link: logger (Killed (carroll.freenode.net (brunner.freenode.net <- dahl.freenode.net))))

19:15:05 Topic now RDF/Semantic Web 24x7 chat - http://rdfig.xmlhack.com/ blog

19:15:05 Users on #rdfig: logger grove ronwalf mancjew DanCon Grantbow Morbus GabeW tim-away danbri em SeeTemp thelsdj attila xower eikeon danbri__ deltab qmacro sandro bwm shellac bijan Mike dmiles pxh|away MarkB

19:15:05 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

19:21:45 <sbp-> sbp- is now known as sbp

19:23:13 <so14k--> [GlobalNotice] Hi all, we are experiencing some server problems, they should be resolved shortly, thanks

19:46:15 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

20:03:23 <danbri> fyi, rdfweb.org and xmlns.com now offline (server move)

20:03:32 <danbri> (in case anyone cares :)

20:03:57 <sbp> when are you estimating that they'll be back up?

20:04:11 <danbri> i'm not

20:04:20 <danbri> by time dns has propagated, hopefully

20:04:27 <libby> sysadmins are hard at work in our living room

20:04:43 <sbp> scary

20:04:59 * danbri ..ooOO("what does this button do?")

20:05:12 <sbp> heh

20:11:27 <niq> did you brisl folks have fun with your power supplies?

20:11:43 <niq> (bbc thinks you did)

20:16:00 <libby> we were down for about 1/2 hour last night

20:17:30 <danbri> niq, got the bbc reference?

20:18:08 <niq> radio 4 talking about alleged storms

20:18:22 <libby> a bloke from round our way sent this around: http://home.btclick.com/litebase/blackout/

20:19:08 <libby> he siad everyone cheered where he was when it came back on

20:19:30 <niq> half hour? We routinely get longer cuts than that, and it ain't always in "interesting" weather

20:19:55 <libby> really?

20:20:22 <libby> it was awful!

20:21:07 <niq> what, the powercut?

20:21:22 <niq> .me thought you lot were all on laptops:-)

20:23:55 <libby> yeah, but our firewall - and so network- went down

21:44:48 <heflin> Hi Tim. You wanted to talk about imports. Is now a good time?

21:53:47 <tim-away> Hi -

21:53:59 <tim-away> heflin?

21:54:16 <tim-away> I was interested in understandingwhat you needed them for.

21:57:03 <tim-away> (Then wasn't a good time, now might be ;-)

21:59:04 <heflin> I'm here. I was at my other computer

22:00:01 <tim-away> I have that problem too.

22:00:29 <tim-away> Anyway, I understand you were promoting daml:imports and I was a bit nervous about it and wondered what you needed it for.

22:00:29 <heflin> So you want to know why I think imports is critical?

22:00:36 <tim-away> yes

22:00:45 <heflin> Here's a simple use case:

22:01:01 <heflin> Imagine I am using OWL for e-commerce contracts.

22:01:15 <tim-away> ok

22:01:23 <heflin> Now as the contractor, I might specify certain conditions in an ontology.

22:01:44 <heflin> I might also have certain conditions that apply to all of my contracts, and put these in a different ontology.

22:01:55 <tim-away> yes.

22:01:59 <tim-away> tim-away is now known as timbl

22:02:16 <heflin> Now, your bidding agent must decide whether my contracts is acceptable to it.

22:02:35 <timbl> Yes.

22:02:38 <heflin> If it does not know which set of ontologies apply, it cannot make this decision.

22:02:55 <heflin> One way of specifying the relevant ontologies is to use imports.

22:03:16 <heflin> In my contract, I import the relevant ontology, which may import more relevant ontologies.

22:03:40 <heflin> The meaning of the contract is defined by the collection of these ontologies.

22:03:46 <timbl> Well, presumably you have different namespaces for the two ontologies.

22:03:58 <timbl> my:lifetimeGuarantee

22:04:10 <heflin> Possibly, but not necessarily.

22:04:12 <timbl> massachusetts:mimimalLegalGuarantee

22:04:29 <heflin> The ontologies might consist entirely of rules that use concepts from other namespaces.

22:04:33 <timbl> You said ytou had two different ontologies.

22:05:00 <heflin> Yes. An ontology is more than just a set of symbols.

22:05:23 <heflin> It is also definitions for those symbols (axioms).

22:05:58 <heflin> Keep in mind, that RDF allows you to say anything about anything.

22:06:05 <timbl> yes.

22:06:29 <heflin> Thus, in one schema (ontology), I can't make statements about symbols (URIs) in other schemas.

22:06:54 <heflin> I don't have to create any new IDs to have a new schema. I can simply add to the axioms of someone elses schemas.

22:07:14 * timbl buzzzzzt

22:07:29 <heflin> I don't understand buzzzzt?

22:08:09 <Grantbow> lol

22:08:10 <timbl> hang on ... you said "RDF allows you to say anything about anything" and then immediately "I can't make statements about symbols (URIs) in other schemas".

22:08:49 <heflin> Whoops. I meant "I can make..." (s/can't/can)

22:08:58 * timbl bzzzt meant "stop there and explain"

22:09:05 <timbl> Ahhh

22:09:17 * timbl resets the bzzzeer

22:09:32 <heflin> Thanks for catching the typo. Anyhoo...

22:09:47 <timbl> You say you can add to the axioms of someone else's schema.

22:10:02 <timbl> That you *can* technically do with RDF. Socially,...

22:10:22 <heflin> Yes. But the problem is what does it mean?

22:10:28 <timbl> ... if it is their term you would be wrong to say things which they did not agree with.

22:10:34 <timbl> Maybe we need some examples.

22:11:07 <heflin> Presumably you would be wrong. But how can you prevent such things.

22:11:23 <heflin> Here's the fundamental problem:

22:11:30 <timbl> You can never prevent people from lyingbut you can avoid listenbing to them.

22:11:52 <heflin> in RDF you can say anthing about anything, but presumably only the owner of the URI can determine its meaning.

22:12:15 <heflin> Therefore, you can say anything, but then it doesn't have any meaning.

22:12:25 <heflin> Or else, the owner can't fix the meaning.

22:12:57 <timbl> When you are processing data, theer is a certain set of documents which you use as input.

22:13:01 <heflin> One way to fix the meaning is to explicitly say what other statements from what other document go along with it.

22:13:22 <heflin> But we can't assume a certain set of documents as input on the Web.

22:13:34 <timbl> Why not?

22:13:59 <heflin> If my agent comes across your contract by randomly searching the Web, how does it know which ontologies it agrees to?

22:14:19 <heflin> If I assume the documents I have seen so far, this might be different than the set you use.

22:14:45 <timbl> The protocols stack and the social agreements lend a special significance to a document you get by dereferencing a term.

22:15:08 <heflin> That just went right over my ahead. Could you explain?

22:15:14 <heflin> s/ahead/head

22:15:26 <timbl> Therefore an interesting set is the set of documents you getby recursively dereferencing all terms.

22:15:36 * timbl backs up 1 line

22:16:02 <timbl> If I allocate a URI in spaec I control - like http://www.w3.org/2002/10/30-whatever

22:16:06 <heflin> Yes, that is in interesting set but not necessarily the only relevant set.

22:16:28 <heflin> Go on...

22:16:38 <timbl> ... then the HTTP spec gives me the power to control what a server gives you as representation of that thing.

22:16:54 <heflin> yes

22:17:03 <timbl> Then any statement I make on that page you can show was authorized by me.

22:17:12 <heflin> with you so far

22:17:26 <timbl> To the extent that (a) I can be bothered and (b) I have the langauges to do it, I can describe what I mean by th eterm in that page.

22:18:03 <heflin> Here, you assume that you describe this entirely at that page?

22:18:17 <timbl> So all documents are not equal, when we talk about foo#bar -- foo can be shown (in http space) to be an authoritative document.

22:19:13 <heflin> Sure, if everything that is relevant is defined at that URIs that correspond to the namespaces, then yes.

22:19:17 <timbl> entirely? Well, that becomes a philosophical issue. I will probably give all the RDFS and OWL constarints on the terms, and a summary in english. But eth eenglish and maybe the rDF will refer indirectly to other stuff - laws etc

22:20:05 <heflin> But what if you define some contractural rules that only combine terms from other documents in novel ways. How does an application find this document?

22:20:14 <timbl> The stuff at that URL is the stuff which can simply, wthout outsdie agreement, be said to be agreed as common understanding by sender and receiver of the original document.

22:20:42 <heflin> This is a very constraining definition.

22:21:20 <timbl> You are always free to make agreements just by pointing to them explicitly in an out-of-band arrangement of course.

22:21:20 <heflin> The example I was giving depends on the ability to arbitrary statements distributed across many pages.

22:21:30 <timbl> Why is it constraining?

22:22:08 <heflin> (back up 1 line): Ah-ha. So imports is essentially taking the out-of-band arrangement and putting it in band. So that agents can use it.

22:23:05 <heflin> It is constraining b/c it doesn't let you point to other documents that might be relevant.

22:23:16 <timbl> Well, I don't know what imports is ;-) You are telling me...

22:23:38 <heflin> That's what I'm telling you.

22:24:04 <timbl> In fact, when you write your ontology, any term you use from anotherr ontolgy will point to it, for the purposes of anyone really trying to find all the documents involved. So you can end up with quite a lot of documents!

22:24:27 <heflin> Instead of relying on out-of-band arrangements (I call you up and say use these ontologies to reason about my documents), imports allows you to specify this in your OWL file.

22:24:36 <timbl> Yes.

22:24:52 <heflin> Do you see the value of that?

22:25:06 <timbl> I do see the value of asserting the contents of anoerth document.

22:25:24 <timbl> BUT actually what you will want to do is to make some statement which brings the document in.

22:25:26 <heflin> I once said imports is like an assertional hyperlink

22:25:43 <timbl> It's not just a questiuon fo saying "the Universal Common Code is true".

22:25:58 <heflin> What do you mean by "bring the document in?

22:26:13 <timbl> You need to say "This contract is a bindingContract under the Universal Commmercial Code".

22:26:21 <heflin> okay

22:26:46 <timbl> When you do that, you lmake a semantic link which gives the relationship of the business at hand to the stuff in question.

22:27:19 <heflin> I think I'm following you...

22:27:20 <timbl> And bingo! the UCC document is hooked in as relevant,and the contract has type ucc:bindingContarct and all itys superclasses.

22:27:55 <heflin> And how would this work without imports?

22:28:17 <timbl> Just like that.

22:28:28 <timbl> <> a ucc:bindingContract. # done

22:29:09 <timbl> @prefix ucc: <http://www.comrce.gov.usa/worldinrdf/uycc.daml>.

22:29:24 <heflin> For something simplistic like this, I guess that would work.

22:29:29 <timbl> :)

22:30:15 <timbl> I am not saying that imports is a bad thing.

22:30:20 <heflin> Maybe I need to try my argument from a different angle, while I chew on this.

22:30:52 <heflin> I understand.

22:31:33 <timbl> I am saying that the functionality is useful but here are many ways of doing it and you need a new model theory for the web. I can write axioms for daml:imports using my log:semantics and log:includes. But i think the community would wantto discuss al this. So it shouldn't be core webont.

22:31:59 <heflin> Essentially, you have a proposed a work-around to the problem I presented that requires me to define things in a special way, so that a default policy of merging all used namespaces causes you to get all relevant information.

22:32:34 <timbl> I would love log:semantics to be standarized. (It is the relationship between a URI and an RDF/N3 formula which is a valid represnetation fo it parsed)

22:32:48 <heflin> The problem is, some people may have different axiomatizations of the same common terms.

22:33:23 <timbl> Yes.

22:33:31 <timbl> So I don't think it can be just slipepd in

22:34:03 <heflin> i.e., there is a namespace with few axioms, and then multiple schemas that don't declare new names but list axioms for the first namespace.

22:34:21 <heflin> There is no way to reference these schemas in RDF b/c you don't use any names from them.

22:35:00 <heflin> I'm not suggesting that we just slip imports in. I think it is a fundamental issue.

22:36:37 <heflin> Ontologies are not the only thing you might like to import. Any document could pertain to the conlcusions that you want drawn from you document.

22:37:06 <timbl> So the original schema ought to put a link the new rules. Yes, I'd like that. I'd like specifically in fact a pointer which lets you knwo that the object is a set of rules - so a rule-oriented system would follow tem.

22:38:00 <timbl> By the way, for your use, presumably a class would do as well as a property?

22:38:14 <heflin> As well as property for what?

22:38:45 <timbl> Instead of <> daml:imports foo.edf you would say foo.rdf rdf:type daml:TrueDocument.

22:38:54 <timbl> for daml:imports

22:39:40 <timbl> It is really a unary function not a binary one I think.

22:39:41 <heflin> I'm not wild about that idea. When you merge documents, you lose track of true from the perspective of what document?

22:40:28 <heflin> It is technically a relationship between two documents which is why a property is better. However, this has its own problems, as the OWL WG has discussed

22:40:32 <timbl> When you merge two documents, you comine their triples. If you endup with two statements about external thinsg being truedocuments then it works.

22:40:51 <heflin> However, someone might want to do more complex reasoning.

22:41:10 <timbl> If you end up with two statemenst about the pre-merged individtal documents imprting things, it doens't help you unless you rememebr that those documents were ones you were believing.

22:41:27 <heflin> For example, in the future, someone might decide to calculate truth of things by analyzing the imports structure of a set of documents.

22:41:48 <timbl> yes, but inthe future they will be able to look at who said what.

22:43:09 <heflin> Sorry Tim. A student just walked into my office. Can I get back to you?

22:43:27 <timbl> { :doc log:semantics :s. :s log:includes { :doc2 a :TrueDoc }} <=> { :doc :imports :doc2 }

22:43:37 <timbl> Jeff, I have to go anyway

22:43:51 <timbl> Thanks for the time

22:44:23 <heflin> Can we pick this up again another time?

22:44:53 <timbl> Sure .. I'd like to... dunno when ..

22:45:02 <timbl> timbl is now known as tim-away

23:35:51 * niq is away: having a well-earned nervous breakdown

23:41:45 * danbri sighs at hours eaten by sysadmin fiddling, heads off

23:54:59 * niq is back (gone 00:19:08)