Semantic Web Interest Group IRC Chat Logs for 2002-11-26

00:29:38 <molar> hi everyone. now i'm very confused. what is the advantage of storing RDF in a persistent database rather than a relational one?

00:34:25 <Guha_> hi danbri

01:12:59 <sandro> molar, isnt' a relational database presistent?

01:18:50 <wiBijan> sandro: Not necessarily.

01:18:53 <molar> well I was getting confused

01:19:01 <wiBijan> I've used a memory only persistent relational DB

01:19:12 <wiBijan> I think.

01:19:27 <molar> RDFStore uses BerkerleyDB which i've never come across before

01:19:33 <molar> that was confusing me

01:19:44 <wiBijan> But, granted, when people say "store rdf in a persistent db" then mean "anything including most rdbmss"

01:19:59 <wiBijan> BerkerleyDB is used as the substrate of mysql and postgres

01:20:23 <wiBijan> The best reason no to use a full RDBMS, SQL varient is that you typically don't need most of what they have

01:21:46 <wiBijan> And, indeed, the overhead can slow you way down.

01:22:13 <wiBijan> Parka, for example, uses a relational engine as it's substrate, but a custom one to avoid the overhead of, for example, SQL parsing.

01:31:37 <molar> thanks for explaining

02:51:13 <AaronSw> AaronSw is now known as djbbot

02:51:20 <djbbot> djbbot is now known as AaronSw

10:04:27 <JibberJim> On the unprefixed about/ID/resource etc. Is there a link to the discussion on why it's needed, as a rough and ready parser creator it seems a lousy idea at first reading to me.

10:05:32 <kao> wasn't it something about mozilla/adobe compatibility?

10:05:53 <JibberJim> The rdf-syntax-grammar doc seems to say it's purely because of legacy RDF-MS rdf/xml and I don't see why that doc should get special treatment?

10:06:42 <JibberJim> I know there's a lot of it about, and I wouldn't mind allowing parsers to consider them the same (for backwards compatibility), but I don't like requiring it.

10:07:58 <kao> agreed, i had expected parsers to handle backwards compatibility. should not be too difficult.

10:08:51 <JibberJim> my rough and ready parser probably won't do it... (but it's only designed for consuming known RDF in any case)

10:09:29 <kao> in general i'm all pro prefix, for disambiguation

10:09:54 <dajobe> oh not this again ;)

10:10:06 <kao> right, i'll shut up :-)

10:10:09 <dajobe> ;)

10:10:27 <JibberJim> all I asked for was a link to the discussion, not to have the discussion...

10:10:53 <dajobe> carry on discusssing it, but I can't see us (RDF core) changing this *again*

10:11:05 <dajobe> it wasn't much coding, only 5 or 6 terms

10:14:04 <JibberJim> it's basically a legacy problem though is it?

10:14:34 <dajobe> yeah

10:15:10 <JibberJim> so it won't matter if the javascript parser doesn't have it then...

10:15:36 <JibberJim> since it's unlikely to ever see legacy documents, just shiny new ones.

10:15:43 <dajobe> unless it is in mozilla

10:16:10 <dajobe> <RDF:RDF

10:16:10 <dajobe> xmlns:NC="http://home.netscape.com/NC-rdf#"

10:16:10 <dajobe> xmlns:RDF="http://www.w3.org/1999/02/22-rdf-syntax-ns#">

10:16:10 <dajobe> <RDF:Description about="urn:search:category:2">

10:16:14 <sandro> It's funny -- after about two weeks of writing a LOT of xhtml by hand, my brain finally does </p> end tags without me noticing, and my HTML almost always turns up valid.

10:16:18 <dajobe> -- example from my mozilla search.rdf file

10:16:28 <JibberJim> mozilla's got its own internal one though for its rdf stuff...

10:16:59 <JibberJim> you'd be silly to use mine for mozilla internals, more if you want to use the same parser on (the various places it manages to work)

11:17:02 <dajobe> yuk, some moz 'rdf' has bare attributes of their own design

12:28:49 <ramin_> ramin_ is now known as ramin

13:23:20 <ramin> A newbie question, but does N3 have the ability to show XML attributes?

13:24:18 <ramin> I.e. something like <rdfs:seeAlso rdf:resource="http://cs.joensuu.fi/~ramin/foaf.rdf" /> is what in N3?

13:37:54 <sandro> Sure: rdfs:seeAlso <http://cs.joensuu.fi/~ramin/foaf.rdf>

13:38:28 <sandro> which has nothing to do with XML, of course. (None of RDF has anything to do with XML, he says, taking a RATHER extreme position.)

13:38:47 <sandro> (RDF's datatypes relate it to XML a bit.)

13:39:50 <ramin> But that is converted to <rdfs:seeAlso>http://cs.joensuu.fi/~ramin/</rdfs:seeAlso> by CWM...

13:40:12 <ramin> I know, but the XML rep is supposed to be used by software, is it not?

13:40:19 <JibberJim> that would be a bug in CWM though

13:40:36 <JibberJim> since seeAlso takes a resource not a literal doesn't it?

13:43:52 <ramin> I guess it is as its domain and range is a resource

13:44:31 <ramin> So rdfs:seeAlso <...> should be converted into <rdfs:seeAlso rdf:Resource="..."/>?

13:44:44 <JibberJim> I would believe so, but I don't like N3

13:47:08 * ramin is away: My ride calleth...

15:21:03 <DanConn>http://lists.w3.org/Archives/Public/www-archive/2002Nov/att-0092/02-index.html

15:21:04 <dc_rdfig> A: http://lists.w3.org/Archives/Public/www-archive/2002Nov/att-0092/02-index.html from DanConn

15:21:12 <DanConn> A:|XML Schema Datatypes in RDF

15:21:12 <dc_rdfig> Titled item A.

15:21:22 <DanConn> A:Jeremy Carroll 26 November 2002

15:21:23 <dc_rdfig> Added comment A1.

15:39:13 <DanConn>http://xml.resource.org/ietf-analysis/current/draft-etal-ietf-analysis.html

15:39:13 <dc_rdfig> B: http://xml.resource.org/ietf-analysis/current/draft-etal-ietf-analysis.html from DanConn

15:39:22 <DanConn> B:|Toward a Quantitative Analysis of IETF Productivity

15:39:23 <dc_rdfig> Titled item B.

15:39:42 <DanConn> B:Rose, Crocker, 26 Mar 2002

15:39:42 <dc_rdfig> Added comment B1.

15:40:44 <DanConn> B:relates to our work on [formalizing W3C schedule/tracking stuff|http://www.w3.org/2000/01/sw/#calsched]

15:40:44 <dc_rdfig> Added comment B2.

16:37:51 <frRonwalf> frRonwalf is now known as ronwalf

19:44:47 * wiBijan reminds folks that if they're available tonight, 5 or so EST, there's a class full of people who think that the Semantic Web is AI who are *dying* to be disabused of that notion

19:57:38 <jimH> jimH is now known as jhendler

19:58:46 <wiBijan> Hey jim.

21:57:14 <jhendler> Trustbot, golbeck@cs.umd.edu to timbl@w3.org

21:57:14 <TrustBot> No Path from golbeck@cs.umd.edu to timbl@w3.org

21:58:00 <jhendler> Trustbot, bparsia@isr.umd.edu to timbl@w3.org

21:58:02 <TrustBot> bparsia@isr.umd.edu should trust timbl@w3.org at a level 7.582895365504061 on a scale of 1 - 9 according to my calculations.

21:58:16 <jhendler> Trustbot, explain bparsia@isr.umd.edu to timbl@w3.org

21:58:17 <TrustBot> Trust values are a weighted average. In this case, the maximum trust path from bparsia@isr.umd.edu to timbl@w3.org had value 8.0. The minimum trust path had value 6.0. bparsia@isr.umd.edu has 2 neighbors with paths to timbl@w3.org, and the weighted average was 7.582895365504061.

21:58:18 <TrustBot> The maximum length path is bparsia@isr.umd.edu->9.0-> 8.0-> 7.0-> 9.0-> 6.0-> 9.0-> 9.0 -> timbl@w3.org and the minimum length path is bparsia@isr.umd.edu->8.0-> 9.0 -> timbl@w3.org.

21:59:24 * jhendler waves to everyone in the world -- class chat coming up soon

22:00:37 <jhendler> jhendler is now known as SemWebClass

22:00:46 * sethl waves back, is excited to lurk and pick up a few things

22:01:12 * sethl can possibly add real world stories

22:01:51 <SemWebClass> SemWebClass is now known as jhendler

22:02:09 <golbeck> hi

22:02:45 <bparsia_> Any and all sundry folks interested, we're starting class

22:04:04 <sandro> I'm here, although I can't stay that long.

22:04:14 * danbri waves

22:04:16 <jhendler1> jhendler1 is now known as SemWebClass

22:04:21 <SemWebClass> hey sandro.

22:05:06 <SemWebClass> bijan heroically figures out why the chat wasnt projecting - class ready to go...

22:06:06 <SemWebClass> sandro - want to say a few words on thoughts about theorem-proving/logic for web?

22:07:02 * SemWebClass waves at AaronSW

22:07:12 <SemWebClass> hi danbri. How's life in RDF world?

22:07:30 <SemWebClass> (class is getting really boring :-<)

22:07:38 <sandro> Hm, can you be more specific? I think first-order logic is nice, and can be layered on RDF. I think RDF is FOL+datatypes. I think the essential logical hook of RDF is using URIs for logical symbols, which implicitely reaches out to other knowledge bases.

22:07:47 * danbri grins, wonders where to start

22:08:14 <sandro> Err -- I think RDF is a SUBSET of FOL+Datatypes.

22:08:26 <sandro> IRC = TypoLand, and mind races ahead of fingers.

22:08:52 <SemWebClass> class ponders that for a minute

22:09:10 <danbri> Interesting times, basically. Apps that use the basic RDF model are infiltrating real life (via mozilla, adobe, rss, dublin core etc), the current W3C spec work is moving towards completion, and there are loads of people hacking on tools and apps of various kinds.

22:09:37 <SemWebClass> any favorites among the tools you've been seeing?

22:09:53 <SemWebClass> (Class is starting on independent projects - lots of interest in tools)

22:10:35 <danbri> My favourite apps are those that publish and interlink RDF docs in the public Web, and tools that encourage this...

22:10:48 <SemWebClass> sandro, any good examples of web logic tools or demos you think people might ook at.

22:10:51 * danbri tries to avoid shameless self publicism...

22:11:10 <SemWebClass> danbri - we're big on self publicism - feel free to provide pointers

22:11:13 <bparsia_> Sandro, has tim given up on intuitionist logic?

22:12:05 <sandro> No good web logic stuff out there now, no. :-(

22:12:08 <SemWebClass> <Slow connection between Maryland and Boston tonight>

22:12:34 <SemWebClass> seen any good examples of N3 encodings of complex things lately (leading question)?

22:12:50 <danbri> you might look at Damian Steer's BrownSauce RDF browser (java, serverside), at Edd Dumbill's FOAFBot(Python harvester and IRC UI), or at the FOAFNaut (SVG UI) collaboration

22:13:30 <danbri>http://brownsauce.sourceforge.net/

22:13:31 <dc_rdfig> C: http://brownsauce.sourceforge.net/ from danbri

22:13:32 <golbeck> foaf is so great.

22:13:37 <sandro> I don't know if he's really given up in intuitionalism. He needs someone with more authority than me to tell/convince him that it's not an out from paradoxes of self reference. (I've read it helps with Liar, but not with Curry's.)

22:13:38 <golbeck> i love foaf

22:13:40 <danbri> C:|BrownSauce RDF browser

22:13:41 <dc_rdfig> Titled item C.

22:13:53 <bparsia_> It doesn't help with the liar

22:14:01 <bparsia_> Since the strengthened liar kills it dead.

22:14:04 <danbri>http://usefulinc.com/foaf/foafbot

22:14:04 <dc_rdfig> D: http://usefulinc.com/foaf/foafbot from danbri

22:14:16 <danbri> D:|FOAFBot: IRC Community Support Agent

22:14:16 <dc_rdfig> Titled item D.

22:14:29 <danbri>http://foafnaut.org/

22:14:29 <dc_rdfig> E: http://foafnaut.org/ from danbri

22:14:33 <danbri> E:|http://foafnaut.org/

22:14:34 <dc_rdfig> Titled item E.

22:14:48 <danbri> E:|FOAFNaut (SVG interface to harvested RDF)

22:14:48 <dc_rdfig> Titled item E.

22:15:30 <danbri> I've found the FOAF experiments really fun, cos people keep building nifty things on top of it...

22:15:39 <danbri>http://rdfweb.org/foaf/

22:15:39 <dc_rdfig> F: http://rdfweb.org/foaf/ from danbri

22:15:47 <sandro> Anyway, I think some kind of safe truth predicate is necessary, and if we have it he'll be fine with FOL. There seems to be a lot of literature in the last 20 years on such truth predicates; I'm just scraping the surface.

22:15:51 <danbri> F:|FOAF the 'friend of a friend' project

22:15:51 <dc_rdfig> Titled item F.

22:15:55 <danbri> F:|FOAF - the 'friend of a friend' project

22:15:56 <dc_rdfig> Titled item F.

22:16:03 <SemWebClass> (prof is trying to demo foafnaut, but having trouble with slow link)

22:16:15 <danbri> F:See also: [http://xmlns.com/foaf/0.1/|foaf namespace doc]

22:16:15 <dc_rdfig> Added comment F1.

22:16:39 <bparsia_> "Safe" truth predicates? You mean something radically distinct from Tarski?

22:16:43 <danbri> E:See [http://jibbering.com/foaf/foafnaut.svg|SVG URL] (Adobe SVG Plugin needed)

22:16:43 <dc_rdfig> Added comment E1.

22:16:52 <bparsia_> I mean, there's all the liar paradox literature

22:17:02 <bparsia_> but most of them use some sort of stratification.

22:18:00 <SemWebClass> question is asked if anyone knows about tools for generating ontologies from relational database

22:18:54 <sandro> Yes, Tarski made it safe by effectively outlawing it. TIm (and others) think that the real world needs self-reference, and it can be handled. Kripke re-opened this door, Perlis defined it enough for KIF to use it, others have played with various approaches.

22:19:04 <SemWebClass> someone in the class asks "Should we be paying attention to topic maps? How do they compete/compare with this RDF stuff?

22:19:06 <SemWebClass> "

22:19:19 <larsbot> you should be paying attention to topic maps, definitely

22:19:26 <SemWebClass> because?

22:19:33 <larsbot> think of them as a layer on top of RDF that makes it easier to find information

22:19:43 <larsbot> a way of organizing information to make it more findable

22:19:48 <SemWebClass> aren't ontologies (and schemas) that?

22:19:54 <sandro> There's some db stuff in cwm that might dump a Schema. Not sure.

22:20:00 <danbri> they're sort of like RDF itself, and sort of like what an RDF Thesaurus vocabulary is for.

22:20:02 <larsbot> they are, but they are much lower-level than topic maps

22:20:02 <SemWebClass> and has someone hooked topic maps into RDF?

22:20:15 * larsbot has done lots of work on that, pointers availalbe on request

22:20:24 <sandro> you should definitely not be payaing attention to topic maps. :-)

22:20:28 <SemWebClass> requested...

22:20:44 <larsbot> start here: http://www.ontopia.net/topicmaps/materials/tmrdfoildaml.html

22:20:52 <larsbot> it's a bit out of date, but helps

22:20:53 <danbri> for the chump?

22:20:59 <SemWebClass> (somehow the folks teaching the course didn't mention topic maps too much - Jim blames Bijan)

22:21:02 <sandro> (you should let Lars pay attention to that, while you work on the lower layers.)

22:21:02 <larsbot> chump has seen it before :)

22:21:08 <danbri> not a problem!

22:21:14 <danbri>http://www.ontopia.net/topicmaps/materials/tmrdfoildaml.html

22:21:14 <dc_rdfig> G: http://www.ontopia.net/topicmaps/materials/tmrdfoildaml.html from danbri

22:21:19 <SemWebClass>http://www.ontopia.net/topicmaps/materials/tmrdfoildaml.html

22:21:19 <dc_rdfig> H: http://www.ontopia.net/topicmaps/materials/tmrdfoildaml.html from SemWebClass

22:21:25 <SemWebClass> G:| topic maps and RDF

22:21:26 <dc_rdfig> Titled item G.

22:21:44 <SemWebClass> larsbot - will be easier for students to find here than to trace through old chumps...

22:21:45 <danbri> H:|Topic Maps and RDF...

22:21:45 <dc_rdfig> Titled item H.

22:21:49 <danbri> H:...again

22:21:49 <dc_rdfig> Added comment H1.

22:21:50 <bparsia_> Topic Maps, Schmopic Maps

22:21:54 <danbri> H:The topic that never dies ;-)

22:21:54 <dc_rdfig> Added comment H2.

22:22:11 <larsbot> basically, RDF is very similar to UML, and UML doesn't make information easier to organize

22:22:19 <larsbot> topic maps add features on top of that that actually help a lot

22:22:30 <shellac_> or OWL....

22:22:32 <larsbot> you can do TM-like things within pure RDF as well, but it's harder

22:22:38 <SemWebClass> question is raised about why topic maps are thought to be higher level than ontologies (prof taught class something different)

22:22:54 <larsbot> what's the difference between topic maps and ontologies?

22:22:55 <danbri> it sits on the awkward fence that separates library classification, thesauri etc from Ontologies, class hierarchies etc...

22:22:58 <larsbot> as far as I know there is none

22:23:33 <larsbot> danbri: yeah, sort of

22:23:48 * sandro waves goodbye

22:23:55 <larsbot> the paper explains why, essentially

22:24:03 <larsbot> RDF has things with properties

22:24:13 <SemWebClass> visiting faculty members points out addison-wesly book recently published that goes through building topic maps on RDF (XML Topic Maps book)

22:24:28 <larsbot> well

22:24:41 <larsbot> topic maps distinguish between different ways of identifying things, and also between different kinds of properties

22:24:41 <SemWebClass> bye sandro

22:24:46 <larsbot> it adds various things on top of that

22:24:53 <larsbot> so it has all that RDF has, plus some extera

22:25:00 <SemWebClass> danbri, question is asked "what is so awkward about that fence?"

22:25:02 <larsbot> the awkward thing is that you could (in theory) build those with RDF

22:25:06 <SemWebClass> (of danbri)

22:25:11 <danbri> What's awkward... OK...

22:25:37 <bparsia_> "Different ways of identifying things"...? "Different kidns of properties"? RDF doesn't do this?

22:25:38 <danbri> One thing that's fun/challenging/etc about RDF, and the SemWeb community, is that several different communities bumped into RDF and saw something of value...

22:25:52 <bparsia_> Or do you mean they've specifically defined a set of useful property types.

22:25:57 <bparsia_> Oops.

22:26:03 <danbri> but they all think about information rather differently. So we have mailing lists with librarians and logicians swapping mail...

22:26:20 <bparsia_> I see the "In theory" claim too late to not type :)

22:26:41 <danbri> from a library perspective, thesauri are the big thing still; from a kr/ai/onto perspective, thesauri may seem a bit old hat...

22:26:47 <danbri> ...or just bad/vague

22:27:01 <larsbot> they're built into the model in TMs, but not in RDF. therefore using these differences is easier in TMs

22:27:05 <danbri> ontologies say things like 'all Dogs are Mammals'

22:27:23 <danbri> ...so you can conclude from 'fido is of type Dog' that 'fido is of type Mammal'.

22:27:40 <danbri> ie in RDF, fido --type--> Dog --subClassOf--> Mammal

22:27:48 <SemWebClass> JimH wonders if larsbot has actually looked at OWL?

22:27:51 <danbri> whereas lots of thesauri don't draw such distinctions, and just say

22:28:09 <danbri> fido--broaderTerm--> Dog --broaderTerm ---> Mammal etc

22:28:11 <SemWebClass> (but then as chair of the OWL WG, JimH is biased))

22:28:22 <danbri> So the awkwardness is that there are

22:28:26 <danbri> (i) different traditions

22:28:34 <danbri> (ii) different representational conventions

22:28:47 <danbri> (iii) different costs associated with precise vs vague classification

22:28:56 <larsbot> SemWebClass/JimH: I've looked at it, but not very carefully. Seems like an extension of DAML+OIL. Correct?

22:29:05 <danbri> and it can _all_ be done in RDF. Choosing to use RDF is just the beginning of the fun... ;-)

22:29:17 <SemWebClass> basically consider it DAML+OIL for now - adds a few things, fixes a few things, better documents.

22:29:36 <larsbot> SemWebClass: in that case it is pretty much equivalent to what TMCL will do in the TM world

22:29:59 <SemWebClass> but I could ask same re: DAML -- i.e. the things you're saying TMs do are in the coverage (I'm really trying to push for some more specificity in what you think TMs can do at a "higher level")

22:30:50 <larsbot> SemWebClass: I think to give you a full answer would flood the channel

22:31:11 <larsbot> TMs and RDF are very very close, but sufficiently far apart that a merge does not make sense

22:31:15 <SemWebClass> we're using you as a guest lecturer on Topic Maps, feel free to go ahead for a bit...

22:31:27 <larsbot> ok

22:31:32 <SemWebClass> but we're really looking for a "what TMs are really good for is..."

22:31:57 <larsbot> the main strength of topic maps is to make it easier to organize information

22:32:08 <larsbot> this comes in two parts, basically:

22:32:18 <larsbot> 1) a model that makes it easy to build useful information structures

22:32:28 <larsbot> 2) a way of thinking about information that makes information easier to find

22:32:41 <larsbot> of course, 2) can be applied using RDF or an RDBMS as well, but TMs make it easier

22:33:03 <larsbot> the perfect introduction to 1) is really Steve Pepper's "The TAO of topic maps"

22:33:15 <larsbot>http://www.ontopia.net/topicmaps/materials/tao.html

22:33:16 <dc_rdfig> I: http://www.ontopia.net/topicmaps/materials/tao.html from larsbot

22:33:22 <larsbot> I:|The TAO of Topic Maps

22:33:22 <dc_rdfig> Titled item I.

22:33:29 <larsbot> I:if you want to learn TMs, start here

22:33:29 <dc_rdfig> Added comment I1.

22:33:37 <SemWebClass> Lars - following was said in class... would you agree:

22:33:47 <larsbot> the key is splitting the universe into topics, associations, and occurrences

22:33:53 <SemWebClass> If you think of web as "one big book", TMs are the indexing structure.

22:34:01 <larsbot> yes, I would say so

22:34:09 <SemWebClass> If you think of Semantic Web as one big formula, they're less clearly that.

22:34:32 <larsbot> I agree, and this is where I think the SW and TMs have different agendas

22:34:47 <SemWebClass> danbri, still around?

22:34:53 <larsbot> that's not so much about technology as it is about goals, but it does have a tech impact

22:35:01 <larsbot> btw, tell me when you want me to stop :-)

22:35:15 <SemWebClass> (JimH agrees with larsbot) - you're doing fine, think that has been very helpful.

22:35:32 <SemWebClass> Nice to get someone from the TM committee to visit class...

22:35:39 <larsbot> happy to help :-)

22:35:49 <larsbot> if anyone has questions, please do feel free to send email (larsga@ontopia.net)

22:36:12 <larsbot> anyway, thinking of the universe in terms of TAOs really helps organize information

22:36:24 <larsbot> that's the key thing. merging and scope are just nice-to-haves

22:36:48 <SemWebClass> suspect Larsbot and JimH have just stopped agreeing :->

22:37:09 <SemWebClass> other semantic web questions?

22:37:30 <SemWebClass> danbri, there's a tools for RDf question - you still here

22:37:43 * danbri missed it, sorry

22:37:47 * danbri scrolls back

22:37:57 <SemWebClass> no no we haven't asked it yet :->

22:38:21 <SemWebClass> question is geenrally about inferencing engines and things - Triple is mentioned

22:38:26 * danbri just reached the same conclusion

22:38:31 <SemWebClass> is there anything you know of that people are playing with and like?

22:38:43 <SemWebClass> esp. on the Web

22:38:50 <danbri> Some people swear by Cwm for hacking...

22:38:56 <danbri> (...some people swear at... ;-)

22:39:29 <danbri> In Java, there's Jos De Roo's Euler inference engine. Small. Reads N3.

22:39:45 <danbri> I've not looked at Triple. There are also a fair few things that map into Prolog or other KR languages.

22:39:58 <SemWebClass> (we look for cwm and euler pages to chump)

22:40:07 <SemWebClass>http://infomesh.net/2001/cwm/

22:40:08 <dc_rdfig> J: http://infomesh.net/2001/cwm/ from SemWebClass

22:40:13 <danbri> I'm probably the wrong person to ask right now, lost touch with the detail of what's what re inference tools.

22:40:23 <SemWebClass> J:| cwm page

22:40:23 <dc_rdfig> Titled item J.

22:41:05 <danbri> I did initiate a project to embed RDF reasoning within Mozilla. Geoff Chappel did all the work (using SWI-Prolog). It was a couple of years ago, probably worth revisiting (and possible student project material).

22:41:09 <SemWebClass> question is asked as to whether there are any SW tools that could be said to be at "production" level yet -- as opposed to demos and ongoing projects

22:41:22 <danbri>http://www.mozilla.org/rdf/doc/inference.html

22:41:22 <dc_rdfig> K: http://www.mozilla.org/rdf/doc/inference.html from danbri

22:42:00 <SemWebClass> ?

22:42:00 <danbri> K:|Mozilla RDF / Enabling Inference

22:42:00 <dc_rdfig> Titled item K.

22:42:48 * danja wonders scope of 'tools'

22:42:51 <danbri> depends quite what you mean by 'SW'. Adobe ship RDF code inside various things (incl. Photoshop).

22:42:52 <danbri>http://www.adobe.com/products/xmp/main.html

22:42:53 <dc_rdfig> L: http://www.adobe.com/products/xmp/main.html from danbri

22:42:54 <dajobe> if you type 'rdf' into google, you get an advert for Intellidimension who provide production SW tools. Geoff Chappel is one of them

22:43:07 <libby> Jos' Euler is suppose to be v fast indeed

22:43:09 <danbri> L:|Adobe XMP

22:43:10 <dc_rdfig> Titled item L.

22:43:34 <SemWebClass> (JimH says hi to libby and invites her to say some stuff about SQUISH or other RDFDB tools)

22:43:41 <danbri> In the Java world, there seems to be a lot of energy around Jena and its ARP parser.

22:43:54 <sethl> we had great luck with Jena

22:44:25 <shellac_> Yeah - one (big) company is using it very heavily

22:44:38 <SemWebClass> XMLspy advert also appears (free RDF editor)

22:44:44 <SemWebClass> know anything about that?

22:45:00 <sethl> I think it's marketing sping... I don't think they have any specific RDF tools

22:45:06 <sethl> er.. spin

22:45:07 * danbri nods (suspected same)

22:45:18 <danja> free as in lots of $$$

22:45:38 <SemWebClass> (Jena still a bit slow, betting is that he will improve soon as new docs are stable)

22:45:40 <danbri> Protege (the ontology tool) had a fair sized user community in science/academia; that tool uses RDF now.

22:45:49 * SemWebClass he = Jeremy

22:46:13 <dajobe> you are confusing jena and arp there

22:46:17 <libby> hey jim

22:46:17 <SemWebClass> yes, protege also going to support OWL (at least Lite) -

22:46:36 <golbeck> protege also has Jess available within it

22:46:43 <golbeck> and pretty soon will have DAML support

22:46:44 <shellac_> jeremy & chris dollin are doing much of the jena2 work

22:46:46 <golbeck> (or so i'm told)

22:46:53 <danbri> I'd like to learn more about Jess's use with RDF...

22:47:05 <dajobe> there is more than just jena. sesame has very good DB links. KAON has great links to J2EE and business end of things

22:47:17 <libby> what's jess?

22:47:17 <dajobe> sesame also does inference

22:47:27 <SemWebClass> one of the students has done a lot of work with Jess and RDF, we'll have him contact danbri later...

22:47:32 <danbri> jess is a java inference engine

22:47:35 <danbri> cool

22:47:57 <shellac_> jena does (some) inferenceing as well

22:47:59 <SemWebClass> libby - anything new on calendar or query fronts?

22:48:45 <libby> I've been pulluing together notes from a calendar woprkshop we had in bristol: http://www.w3.org/2001/sw/Europe/reports/dev_workshop_report_2/

22:48:53 <libby> - not finished yet but gettign there

22:49:03 <libby> query...not really

22:49:57 <libby> I've resorted to strraightfoward SQl queries in an attempt to analyse the content of my database

22:50:00 <libby>http://swordfish.rdfweb.org/discovery/2001/08/codepict/stats.html

22:50:00 <dc_rdfig> M: http://swordfish.rdfweb.org/discovery/2001/08/codepict/stats.html from libby

22:50:02 <danbri> same old same old? ;)

22:50:38 <libby> ...which indicates that my API needs updating ...which it is being...for provenance, stuff like that

22:50:46 <danbri> The queries you wanted to do included asking about which documents had used which predicates etc., right? That could be pretty hairy to do in pure RDF...

22:51:01 <danbri> ...since most RDF query languages don't seem to expose the 'who said what' aspects for query.

22:51:26 <libby> indeed, thought I'm hoping can get that out of an API (like your ruby one)

22:51:49 <libby> I've gone back to the api/provenace thing because it's essential for merging calendars, for example

22:52:01 <SemWebClass> (Class avoids getting into the context issue)

22:53:12 <danbri> One question I'd turn back to the class is... 'how much inference is enough?'

22:53:30 <SemWebClass> danbri - can you explain?

22:53:52 <danbri> can we built practical apps without using inference/rule systems, or is there a base level of 'smarts' that is both implementable and makes the app-creators life vastly easier?

22:54:05 <danbri> not sure if that helped.

22:54:07 <libby> M:|RDFWEb database stats

22:54:08 <dc_rdfig> Titled item M.

22:54:25 <SemWebClass> helped a lot - comments being solicited ...

22:55:03 <libby>http://rdfweb.org/people/damian/2002/02/foafnation/

22:55:04 <dc_rdfig> N: http://rdfweb.org/people/damian/2002/02/foafnation/ from libby

22:55:14 <libby> N:|Codepiction Paths - Explanation

22:55:15 <dc_rdfig> Titled item N.

22:55:17 <SemWebClass> one student says "without inferencing it's just another search engine" so it's clear we need some...

22:55:44 <libby> N:special case of inferencing for somethign specific - Damian Steer

22:55:44 <dc_rdfig> Added comment N1.

22:55:46 <danbri> ...a search engine that knows how to merge data at a much finer grain, I'd argue.

22:55:50 <SemWebClass> expression by a student that we could get by with a base level for now, doesn't seem like a whole full-scale engine is needed

22:56:08 <SemWebClass> (jimH threatens to put this question on the final exam)

22:56:13 <danbri> ...but a lot of the data-merging capabilities of RDF etc assume that we're all using the same URIs to name entities of common interest.

22:56:15 <libby> M:working on this with Jim Ley but I've forgotten the url he had now :(

22:56:15 <dc_rdfig> Added comment M1.

22:56:28 <SemWebClass> another dtudent says - seems like merging data involves inferencing.

22:56:36 <danbri> In practice, thats not always the case... so imho there's a particular role for inferences that help us identify when we're talking about the same thing.

22:56:55 <danbri> yup... like the owl:InverseFunctionalProperty mechanism, or whatever its called now...

22:57:21 <SemWebClass> question is asked "how shall we quantify inferncing to answer that question"

22:57:26 <danbri> naive rdf data merging says 'just shove all the triples in one big store, and where the URIs are the same, the datasets join up'

22:57:34 <danbri> heh, good question. Dunno.

22:57:37 <SemWebClass> (Bijan gives a short logic lecture)

22:57:44 <SemWebClass> expressivity defined...

22:58:08 <shellac_> is this about inferencing or the 'size' of the semantics?

22:58:31 <SemWebClass> bijan: as one adds logical machinary one gets more things you can inference, at a computational cost (and undecidable soon)

22:58:40 <danbri> about expectations of fancy logic-oriented processing on the part of consuming apps

22:58:59 <SemWebClass> bijan accused of being member of "logic police" - accepts it happily

22:59:12 <SemWebClass> but question is asked "how will we police logic on the semantic web"?

22:59:28 <danbri> ...and how those expectations affect the decisions folk make when they define new vocabularies, ie. on the cost of creating new (sub-)properties vs re-using a 'near fit' existing one

22:59:45 <libby> that comes up a lot

23:00:08 * danbri suposes the answer is something to do with proof checkers, but doesn't attempt a proper answer

23:00:19 <danbri> We'll get Bijan to check the proofs!

23:00:58 <bparsia_> I'm the charter member of the Logic Police

23:01:12 <bparsia_> "Pull over sir, you're expression was inconsistent"

23:01:16 <bparsia_> your even

23:01:22 <SemWebClass> JimH is expressing in class his belief that we need to design SW to allow for inconsistency, mistakes, conflicts, etc.

23:01:44 * danbri agrees

23:01:51 <SemWebClass> and that is a big change from "AI as usual"

23:01:55 <danbri> a simple use case: photo/picture metadata...

23:02:09 <SemWebClass> sort of Semweb : KR :: Web : hypertext

23:02:22 <danbri> often useful to compose a chunk of RDF that captures the 'view of the world according to' some picture (or other artistic, fictional etc work).

23:02:41 <danbri> we're going to write those .rdf files into public www sites, even though we know they're not actually _true_

23:03:26 <SemWebClass> (vacant expressions disappearing from face of class, danbri on a roll...)

23:03:52 * danbri shuts up now

23:04:06 <shellac_> contexts? again?

23:04:08 * SemWebClass no, no, that was a GOOD thing -

23:04:29 * SemWebClass we were losing them before, they like the photo metadata example

23:04:40 <danbri> Heh, OK...

23:05:09 <danbri> So I was wondering if the class knew RDF's origins with PICS, for content filtering apps

23:05:27 <danbri> ...whether they had any thoughts on the impact of RDF/SW/XML to that domain.

23:05:52 * SemWebClass doesn't know about RDFs origins with PICS (I started with MCF)

23:05:56 <danbri> People used to get very cross and worried about PICS (re possiblities of censorship etc) yet relatively few have aired similar concerns regarding SemWeb.

23:06:02 <danbri> RDF has many origins!

23:06:08 <SemWebClass> PICS?

23:06:13 <danbri> Organisationally, within W3C, it built on earlier work for labelling Web pages.

23:06:26 <danbri> The "Platform for Internet Content Selection" (PICS)

23:06:51 * SemWebClass wishes AcroBot was working tonight :->

23:07:02 <danbri> This defined a syntax for simple web page labels, and protocols for talking to servers that contain classifications of web pages

23:07:26 <danbri> eg. to support 'family filter' apps, as well as various collaborative-filtering sorts of system.

23:07:52 <danbri> It got implemented in Internet Explorer, IE, and connects to mainstream concerns about the Internet as a swirling cesspit of filth.

23:07:55 <SemWebClass> AcroBot: PICS: prodigiously indoles centralistic sketching

23:08:16 <danbri>http://www.w3.org/PICS/

23:08:16 <dc_rdfig> O: http://www.w3.org/PICS/ from danbri

23:08:32 <danbri> O:|W3C Platform for Internet Content Selection (PICS)

23:08:33 <dc_rdfig> Titled item O.

23:08:55 * SemWebClass may have to try IE sometime...

23:09:10 <danbri> Hey, I never mentioned Porn!

23:09:14 * danbri on his best behaviour

23:09:23 <SemWebClass> AcroBot: danbri: diathermancy Americanoid neuroplexus bulldozer roentgenize isorhodeose

23:09:52 <danbri> So I was curious about whether SemWebClass sees RDF, SemWeb, Ontology etc as picking up that theme...

23:10:16 <danbri> ...whether controversial content will become harder, or easier, to find...

23:10:16 <bparsia_> WHich theme?

23:10:19 <bparsia_> Ah.

23:11:14 <SemWebClass> bijan tells class about danbri's "rude thing" work in SVG

23:11:56 <danbri> since you insist...

23:12:16 <SemWebClass> we do, we do...

23:12:18 <danbri>http://rdfweb.org/2002/svgsemantics/picsng-demo.html

23:12:18 <dc_rdfig> P: http://rdfweb.org/2002/svgsemantics/picsng-demo.html from danbri

23:12:24 <shellac_> please, not again...

23:12:28 <danbri> P:|SVG Semantics "PICS PG" demo

23:12:28 <dc_rdfig> Titled item P.

23:12:53 <danbri> P:Basic idea... that the selfsame technology that can be used for filtering, can also be used for discovery.

23:12:53 <dc_rdfig> Added comment P1.

23:13:04 <SemWebClass> (class ends soon, any final comments)

23:13:08 * shellac_ unplugs rdfweb.org

23:13:11 <SemWebClass> we thank you all much!

23:13:16 <danbri> heh, thanks

23:13:36 <SemWebClass> larsbot, if you're still here - many thanks for TM thoughts (students had to get them somewhere ;->)

23:14:23 <larsbot> SemWebClass: you're very much welcome :)

23:14:39 <bparsia_> (We burned you in effigy, you TM heathen you)

23:14:52 <larsbot> I'm sure you did :)

23:15:13 <SemWebClass> ciao all - time to calm bijan back down and to dismiss the class...

23:15:20 * golbeck gets bijan upset by the sample image on SVG-Semantics: "PICS NG" Demo

23:15:24 * danbri wonders about pre-announcing questions next time, so we get time to look clever

23:15:24 <libby> bye!

23:15:57 <bparsia_> :)

23:16:07 <larsbot> btw, which university were they from?

23:16:08 <danbri> cu

23:16:57 <golbeck> university of MD, college park

23:17:38 <golbeck> SemWebClass was jim hendler

23:17:45 <larsbot> aha. I see :)

23:17:49 <golbeck> he had hoped to engage us, the class, in asking you question

23:17:51 <golbeck> s

23:18:00 <golbeck> he was kind of disappointed in us :(

23:18:10 <larsbot> too bad

23:18:17 <danbri> awww

23:18:18 <larsbot> so most of the questions were from Jim, or?

23:18:34 <golbeck> there were a few from the class

23:18:50 <golbeck> i think we were all kind of unsure what to ask

23:19:00 <larsbot> can understand that

23:19:09 <golbeck> i mean, our research group talks on #rdfig a lot

23:19:13 <larsbot> might be easier if everyone had their own keyboard

23:19:41 <golbeck> perhaps

23:21:05 <danbri> Oh, Golbeck, can you pass on something I've forgotten... that Jos's Euler inference engine also now available in C#

23:21:10 <danbri> ...for MS .NET fans

23:21:14 <golbeck> sure

23:21:24 <danbri> (Jos couldn't get onto IRC, but was watching from the Web...)

23:22:19 * golbeck adds to the logs

23:22:55 <larsbot>http://www.ontopia.net/topicmaps/explorations.html#Autogeneration

23:22:55 <dc_rdfig> Q: http://www.ontopia.net/topicmaps/explorations.html#Autogeneration from larsbot

23:23:10 <larsbot> Q:|Three presentations about a tool for generating RDF

23:23:11 <dc_rdfig> Titled item Q.

23:23:24 <dajobe>http://www.agfa.com/w3c/euler/

23:23:25 <dc_rdfig> R: http://www.agfa.com/w3c/euler/ from dajobe

23:23:33 <dajobe> R:|Euler proof mechanism, Jos De Roo, AGFA

23:23:34 <dc_rdfig> Titled item R.

23:23:43 <larsbot> Q:Sort of relevant to the questions about tools

23:23:43 <dc_rdfig> Added comment Q1.

23:24:04 <larsbot> Q:Used to generate TMs, but goes via RDF

23:24:04 <dc_rdfig> Added comment Q2.

23:24:06 <dajobe> R:RDF, OWL, Inferencing - in Java and proves this stuff works

23:24:06 <dc_rdfig> Added comment R1.

23:24:39 <libby> P:here's the show with the guy with the rudeThing [http://www.bathfringe.co.uk/cgi-bin/display2002.pl?77|Trevor Stuart, Taboo]

23:24:39 <dc_rdfig> Added comment P2.

23:25:25 <larsbot> Q:Oh, and lots of RudeThings in there... :)

23:25:25 <dc_rdfig> Added comment Q3.

23:25:28 <danbri> that'll teach him to be avant garde

23:25:37 <golbeck> hahahahaha

23:25:41 <libby> quite right

23:25:45 <golbeck> ok. class over. time to go home.

23:25:52 <golbeck> thanks again from us for helping :)

23:26:20 * dajobe reminds people - XML Europe 2003 is in London this year, paper abstracts by Dec 15

23:26:55 <dajobe> er 13th

23:44:23 * DanConn wanders by...

23:44:33 * DanConn seems to have missed the party

23:50:42 <AaronSw> Oops, me too; I fell asleep.

23:50:47 <AaronSw> Doesn't seem like I would have been too helpful.

23:50:51 <AaronSw> I would have just gotten caught by the Logic Police.

23:51:06 * AaronSw reads TAO of TM

23:51:54 * AaronSw is wary of specs that cannot be explained on one foot

23:53:42 <AaronSw> Hm, topic maps make sense if I think of them as indexing books

23:55:39 * DanConn wonders if RDF passes the one foot test still/anymore

23:57:12 <AaronSw> not really, so I just most of those rdfcore decisions :)

23:57:40 <DanConn> s/just/just ignore/?

23:58:06 <AaronSw> oops, yeah

23:58:58 <sethl> what is the one foot test?

23:59:15 <DanConn> stand on one foot...

23:59:21 <AaronSw> and explain X

23:59:33 <AaronSw> without falling over

23:59:46 <sethl> gotcha