Semantic Web Interest Group IRC Chat Logs for 2003-10-17

00:07:23 <mdupont> mdupont is now known as md-bk-0800CET

03:34:55 <Mutiny_> Mutiny_ is now known as Mutiny

06:45:33 <D[a]vey> D[a]vey is now known as Davey

07:37:51 <Davey> Davey is now known as D[a]vey

08:58:50 <LeoS> <irc://LeoS@irc.freenode.net/#rdfig> <http://www.w3.org/2002/07/owl#sameAs> <mailto:leo@gnowsis.com>.

09:01:56 <libby> a man who speaks entirely in ntriples!

09:05:26 <LeoS> <mailto:leo@gnowsis.com> <http://xmlns.com/foaf/0.1/interest> <http://www.w3.org/2001/sw/>, <http://www.gnowsis.com>.

09:06:00 <LeoS> <mailto:leo@gnowsis.com> <http://xmlns.com/foaf/0.1/currentProject> <http://www.gnowsis.com>.

09:06:56 <LeoS> <http://www.gnowsis.com> <http://purl.org/dc/elements/1.1/subject> "A personal Semantic Desktop".

09:07:50 <arnarl> hi

09:09:11 * LeoS writes in n3 and uses a brain that is weaved like a web...

10:00:04 <Wack> dajobe! about your question regarding the redland python code on my http space the other day; it was something old that should have been deleted :]

10:00:15 <dajobe> lol

10:00:23 * dajobe customers--

10:01:06 <Wack> I use that webserver to put stuff on I want to show to people, I always forget to remove it afterwards, so it tends to fill up with random stuff

10:02:24 <Wack> still planning to create a rdf based weblog/wiki like repository of remind-self-notes and stuff like this

10:03:06 <Wack> but alas, still too busy with other stuff, just finished over 60 illustrations that go in a book of a client of ours

10:03:17 <dajobe> that's synactics

10:03:46 <Wack> no, synantics is where I work :]

10:04:03 * dajobe looks again at http://www.xs4all.nl/~wackin/2003-10-14_gui-voorstel_revisie_5.png

10:04:19 <Wack> feedback? ;)

10:04:34 <dajobe> <Wack> it's kinda purpose-built for our own rdf storage, but the internals should be able to wrap around redland pretty easily

10:04:55 <dajobe> <Wack> ... (schema aware editing is 'the only way to fly'; we're working on a wxPython editor and plan to implement schema awareness as well)

10:05:01 * dajobe reviews

10:05:36 <dajobe> so it's primarily an editor

10:05:42 <Wack> to a certain degree most rdf api's are alike, ours is heavily based on jena and others, redland can easily be wrapped with the same api as ours

10:06:06 <dajobe> Java/jena with wxPython ?

10:06:11 <Wack> yes, we've got this behemoth of a website framework that stores all it's data in a rdf db

10:06:12 <dajobe> jython?

10:06:58 <dajobe> or is the API based on jena but in python?

10:07:02 <dajobe> I'm confused

10:07:09 <Wack> no, the rdfdb is written in c++

10:07:35 <Wack> it exposes its API thru corba

10:07:54 <Wack> and thus can be used by all languages/environments that can 'talk' corba

10:07:58 <dajobe> ah

10:08:10 <dajobe> this sounds familiar, did we discuss this before?

10:08:52 <dajobe> I considered c++ & corba when I started redland

10:08:53 <Wack> I've mentioned some stuff here before yes

10:09:06 <dajobe> but dismised it since it hadn't got decent perl, python etc. support

10:09:21 <dajobe> anywya, you can always add a C++ wrapper later

10:09:23 <Wack> perl support is sketchy (gnome's orbit)

10:09:51 <Wack> but python has the best corba 'orb' I've seen for a scripting language (omnipy)

10:10:04 <dajobe> I want to make some sort of gnome-friendly API, so it's on my virtual todo

10:10:30 <dajobe> so you have (RDF db in C++ ) <-CORBA-> (wxPython editor in Python/C)

10:10:49 <Wack> basically

10:11:02 <dajobe> apart from convincing you to maybe use redland...

10:11:07 <Wack> :]

10:11:09 <dajobe> ... my other thoughts are on abstract APIs

10:11:19 <dajobe> so if you could make the IDL you use for the corba public

10:11:24 <dajobe> that would be interesting

10:11:50 <Wack> hang on, I'll put some stuff up

10:12:52 <Wack> we've made some speed/overhead concessions when passing back nodes to the db

10:13:22 <Wack> passing object references is rather costly in corba

10:14:18 <dajobe> yes, that's what I though

10:14:27 <Wack> but a string is cheap

10:14:42 <Wack> so we use enum's when the db accepts nodes

10:15:10 <Wack> that can contain both existing objectrefs as well as a string (URI or literal value)

10:15:43 <dajobe> BLURB:CORBA APIs

10:15:44 <dc_rdfig> A: CORBA APIs from dajobe

10:15:50 <dajobe> logger, ptr?

10:15:50 <dajobe> I'm logging. Sorry, searching removed.

10:15:53 <dajobe> hah

10:16:00 <dajobe> hoisted on my own petard

10:16:05 <dajobe> logger, pointer

10:16:05 <dajobe> See http://ilrt.org/discovery/chatlogs/rdfig/2003-10-17#T10-16-05

10:16:14 <dajobe> A:[http://ilrt.org/discovery/chatlogs/rdfig/2003-10-17#T10-16-05|discussion]

10:16:15 <dc_rdfig> Added comment A1.

10:16:39 <dajobe> of course mozilla uses IDLs

10:16:48 <dajobe> which is the first RDF api pretty much

10:17:39 <dajobe> A:[http://lxr.mozilla.org/mozilla/source/rdf/base/idl/|Mozilla's RDF IDLs] using their ORB technology

10:17:39 <dc_rdfig> Added comment A2.

10:18:22 <dajobe> not surprising, both jena and redland were both inspired by that and earlier stuff such as SiRPAC

10:27:04 <Wack>http://www.angelite.nl

10:27:04 <dc_rdfig> B: http://www.angelite.nl from Wack

10:27:06 <Wack> err

10:27:16 <Wack> B:=http://www.angelite.nl/aquamarine/

10:27:16 <dc_rdfig> Replaced URL of B.

10:28:29 <Wack> A:|Doxygen generated documentation of our CORBA RDF database 'Aquamarine'

10:28:30 <dc_rdfig> Titled item A.

10:29:41 <Wack> A: It's part of a website framework 'Angelite' we're developing, all data that drives the website (both static and dynamic) is RDF in Aquamarine.

10:29:41 <dc_rdfig> Added comment A3.

10:29:52 <Wack> err

10:30:00 <Wack> B:|Doxygen generated documentation of our CORBA RDF database 'Aquamarine'

10:30:00 <dc_rdfig> Titled item B.

10:30:22 <Wack> B: It's part of a website framework 'Angelite' we're developing, all data that drives the website (both static and dynamic) is RDF in Aquamarine.

10:30:22 <dc_rdfig> Added comment B1.

10:31:23 <Wack> A1:[http://ilrt.org/discovery/chatlogs/rdfig/2003-10-17#T10-16-05|discussion]

10:31:23 <dc_rdfig> Replaced comment A1.

10:31:31 <Wack> A2:[http://lxr.mozilla.org/mozilla/source/rdf/base/idl/|Mozilla's RDF IDLs] using their ORB technology

10:31:31 <dc_rdfig> Replaced comment A2.

10:31:55 <Wack> A3:""

10:31:55 <dc_rdfig> Deleted comment A3.

10:32:14 <Wack> A:|CORBA APIs

10:32:15 <dc_rdfig> Titled item A.

10:32:51 <Wack> phew fixed :]

10:33:36 <Wack> dajobe: can you read the URL I posted? we've got some freaky dns issue here with the angelite.nl domain at the office, so I cannot verify it works except with lynx on a remote host

10:34:18 * dajobe looks

10:34:22 <dajobe> works

10:36:39 <Wack> B:A good place to start browsing the API is from the [http://www.angelite.nl/aquamarine/interfaceaquamarine_1_1Server.html|Server] and/or [http://www.angelite.nl/aquamarine/interfaceaquamarine_1_1Model.html|Model] interface.

10:36:39 <dc_rdfig> Added comment B2.

10:38:04 <dajobe> hmm

10:38:07 <dajobe> your literals can be structs

10:38:11 <dajobe> and complex objcets

10:38:13 <dajobe> and your properties also

10:38:20 <dajobe> in rdf, properties can only be URIs

10:39:08 <Wack> B:Please note the ability to make 'volatile' statements; these will disappear after the model has been closed, more or less a pseudo-transactional system to facilitate request-specific output in the website framework it is part of.

10:39:08 <dc_rdfig> Added comment B3.

10:39:21 <dajobe> not all models will be able to give you Model.size()

10:39:37 <Wack> yes, it's only syntactic sugar; the namespace & localname parts are joined internally

10:41:23 <dajobe> B:[http://www.angelite.nl/aquamarine/aquamarine_8idl-source.html|aquamarine.idl]

10:41:23 <dc_rdfig> Added comment B4.

10:42:07 <dajobe> where do blank nodes appear?

10:42:13 <Wack> B:We're still considering cleaning up a lot of stuff to make the DB more 'RDF compliant' internally and externally (dajobe noted properties being split in namespace & localname) and other leftovers like statements being resources etc.

10:42:13 <dc_rdfig> Added comment B5.

10:42:30 <Wack> 'anonymous resources'

10:42:31 <dajobe> it does look a little rdf-99 ish

10:42:41 <dajobe> ah yes

10:42:49 <dajobe> using id, not uri

10:43:15 <dajobe> I see the jena influence too

10:45:25 <dajobe> I think jena2 changed the node/resource/literal class relationship IIRC

10:50:07 <Wack> biggest drawback of corba is it's synchronous nature

10:50:28 <Wack> lessons have been learned :]

10:50:34 <dajobe> do tell

10:51:43 <Wack> well, corba is pretty fast when transmitting marshable 'flat' data by value

10:52:11 <Wack> but methodcalls are roundtrip, so latency plays a major role when doing large amounts of calls

10:52:31 <dajobe> that's like any distributed system

10:52:43 <Wack> you're better off being able to wrap a large resultset in a sequence of structs instead of an iterator over objects

10:53:07 <dajobe> that's your DataStream

10:53:15 <dajobe> no, that's octets

10:53:24 <Wack> no, the NodeIterator interface

10:53:37 <dajobe> it's really a NodeSequence

10:54:06 <dajobe> with an iterator api

10:54:51 <dajobe> hmm, you don't have what's typically called, "triplesMatching"

10:55:06 <dajobe> oh right, triples are nodes

10:56:01 <dajobe>http://www.w3.org/Submission/2003/SUBM-rdf-netapi-20031002/

10:56:01 <dc_rdfig> C: http://www.w3.org/Submission/2003/SUBM-rdf-netapi-20031002/ from dajobe

10:56:03 <Wack> triplesMatching? like 'wildcard' searches; <urn:foo> <?> <?>?

10:56:06 <dajobe> C:|RDF Net API submission

10:56:07 <dc_rdfig> Titled item C.

10:56:07 <dajobe> yes

10:56:13 <Wack> Model::find() can do that

10:56:28 <dajobe> chrisc: StatementSet here seems to match NodeSequence in aquamarine

10:56:30 <dajobe> argh

10:56:33 <dajobe> C:StatementSet here seems to match NodeSequence in aquamarine

10:56:33 <dc_rdfig> Added comment C1.

10:57:05 <dajobe> C:at least when aquamarine.Node is a Statement

10:57:06 <dc_rdfig> Added comment C2.

10:57:30 <dajobe> yes, I saw find

10:59:07 <dajobe> I guess for me, I'd expect to have both the iterator api and sequence of nodes

10:59:50 <Wack> we end up using NodeIterator::rest() a lot

11:00:02 <Wack> it returns a corba sequence of Node's

11:02:01 <Wack> but, if you're only interested in the first 3 nodes in a 100 node iterator you can either use the 100% corba hasNext() / next() methods or nextSeq() if you're smart

11:04:14 <dajobe> won't it suck all the data across before you call it?

11:04:39 <dajobe> i.e. the methods that return a NodeIterator - won't they have to make the seqNode full of nodes

11:05:04 <Wack> the seqNode is not a property

11:05:10 <dajobe> I know

11:05:47 <dajobe> oops, I read that wrong

11:06:09 <dajobe> I read the typedef as a data member

11:09:53 <dajobe> I guess this tension between a programatic api and one with more consideration of distribution; you need more "clumping" of frequent ops for effienceny. Similarly for remote queries, you sometimes just find it quicker to pull all the resultset back.

11:20:10 <Wack> yes, well, usually a rdfql could do that

11:26:30 <dajobe> that's actually what some of them do now

11:26:44 <dajobe> taht is; always pull the entire resultset back

11:27:07 <dajobe> so that when you navigate by rows, it's quicker

11:27:16 <dajobe> which is the most common next step

11:35:11 <dajobe> bbl

11:50:39 <md-bk-0800CET> md-bk-0800CET is now known as mdupont

12:16:45 <monkeyiq> is there a public "bookmarks" / links section to interesting owl / rdf files in the rdfig / w3.org site?

12:17:50 * monkeyiq is looking for CIA factbook / wordnet again

12:35:55 <dajobe> monkeyiq: it's off daml.org data somewhere

12:36:05 <D[a]vey> D[a]vey is now known as Davey

12:40:19 <monkeyiq> dajobe: I found this page http://www.daml.org/2003/09/factbook/

12:40:28 <dajobe> that's it

12:40:54 <monkeyiq> but apart from the USA info, there doesnt seem to be a "complete" tarball

12:42:09 <libby> daml.org/data?

12:42:13 <monkeyiq> I guess I could just release a snake on http://www.daml.org/2003/09/factbook/countries.html

12:42:50 <dajobe> you can always just download and use the program linked from there

12:43:21 <monkeyiq> 403 is given for http://www.daml.org/2003/09/factbook/gendaml.java

12:43:42 <dajobe> google cache has it

12:44:16 <monkeyiq> :) I should add that to my download context menu (dl from gcache)

12:55:01 <monkeyiq> 403 city today http://www.daml.org/2001/10/html/airport.java

13:07:19 <timbl-fl> { x1 a A; p B} => { x a C }.

13:07:19 * timbl-fl ooops

13:09:31 <JibberJim> monkeyiq, if you want airport data, I've got some more data than the daml people, and someone is actually cleaning it up even further and anding more stuff from various sources.

13:11:06 <monkeyiq> JibberJim: that would be very handy. is there a website someplace for it?

13:11:44 <JibberJim> - http://jibbering.com/foaf/airports.1?LHR;DFW;ETC

13:13:07 <JibberJim> If you want the whole data set (in sql not rdf) then you can get the sql from /foaf/dumps/ "oPless" is currently cleaning the data, so might be worth seeing him when he's next about.

13:15:05 <monkeyiq> JibberJim: ah ok. I'll grab the SQL version for now.

13:16:02 <monkeyiq> JibberJim: when I saw the airport stuff on the daml site I tought about having it around for looking at air booking sites. type ahead for airport code would be very nice :)

13:17:12 <JibberJim> - http://jibbering.com/flightsearch/ is my start of an airline booking site on RDF/SVG ...

13:17:54 <JibberJim> Although it looks like the backend's been upgraded and it no longer gets prices.

13:18:28 <monkeyiq> JibberJim: I am still combing for little chunks of RDF that I can use in my browser... being able to match some form feilds on pages and perform extra work for me will be a handy thing

13:18:55 <JibberJim> - http://jibbering.com/flightsearch/extract.1?airline=ryanair still works though.

13:19:39 <monkeyiq> virgblue doesn't work from http://jibbering.com/flightsearch/

13:19:58 <JibberJim> and http://jibbering.com/flightsearch/interface.svg for the SVG version.

13:20:27 <JibberJim> Yeah it looks like my scraping of their sites is now broken, they must've changed their HTML.

13:20:58 <monkeyiq> hmm, I'll have to work on getting the acrosvg plugin working again :(

13:21:28 <monkeyiq> seems to view okish in eog.

13:22:27 <monkeyiq> I might stick to just form hinting for starters. I tend to get bitten too much with website scraping stuff.

13:22:39 <monkeyiq> mainly when I make it and the other folks change things in subtle ways

13:25:27 <monkeyiq> though in a way my form thing is a form of scraping... a URL regex + form field regex to attach semantics to the input text

13:43:57 <DanC_jam> can anybody else get to http://esw.w3.org/topic/ ?

13:44:37 <dajobe> DanC_jam: yes

13:49:28 <jh-GoYanks> DanC_jam - how's Florida?

13:49:53 * DanC_jam doesn't speak to yankees fans

13:50:00 * jh-GoYanks gets esw...topic just fine

13:50:17 * jh-GoYanks wishes he'd known that a long time ago :->

13:50:38 <jh-GoYanks> jh-GoYanks is now known as JimH_

13:51:00 <DanC_jam> I'm a royals fan, a sworn enemy of the yankees. (http://dm93.org/1967/6712kc/)

13:51:26 <JimH_> lots of teams are jealous...

13:52:01 <DanC_jam> jealousy is a fine rationale for animosity, no?

13:54:28 <JimH_> yup - I grew up near Shea, but never really got into the Mets, made for lots of fights when I was growing up :->

13:54:46 <DanC_jam> I'm sorta in denial about being in Florida, Jim, since my family isn't here.

13:55:03 <JimH_> (of course, I did root for the Mets in 1986, when they did to the Sox what the Yanks did last night...)

13:55:40 <JimH_> DanC - understood - I hate leaving my kid for the few days I'll be there, and she's a lot older than yours

14:00:58 <timbl-fl> What sis the sate foeth art when it coimes to transforming OWL into rules?

14:01:13 * timbl-fl denial++

14:01:52 <JimH_> OWL into rules? couple different things going on there -- do you mean who has an axiomization of OWL in rules, or do you mean using OWL to represent rules?

14:03:01 * JimH_ timbl-fl seems to be doing a lot more travel then he used to - /me sympathizes

14:08:35 <timbl-fl> I mean using an OWL axiomotization to convert an arbitrary DLP OWL into rules.

14:09:28 <timbl-fl> I understand that the DLP paper had the alogorithm in it. That maskes an interesting langauge the DLP-OWL with rules extenstions, as it can be expressed totally in rukes.

14:09:49 <libby> JimH, meet JimH_ :)

14:09:58 <JimH> hi Jim...

14:10:01 <sbp> oh man

14:10:08 <libby> weird huh?

14:10:26 <sbp> thanks libby. as if having seven Dans wasn't bad enough :-)

14:10:31 * JimH mutters "it's that Jibber bloke and Libby, they made me do it..."

14:10:31 * dajobe identifies the strange w3c link checker requests; somebody was validating http://www.w3.org/2003/03/rdfqr-tests/

14:10:43 <libby> heheh

14:14:29 <JimH_> wondered where my nick was...

15:17:50 <mdupont> mdupont is now known as md-afk

15:40:42 <Davey> Davey is now known as D[a]vey

17:06:01 <danja> no danbri? huh. Hi bitsko - I was going to point him to Sam's blog

17:06:17 <bitsko> :)

17:08:24 <danja> how's your RQL?

17:08:44 <bitsko> mostly inferred

17:08:50 <danja> ;-)

17:08:55 <bitsko> is that the SQL-like one?

17:09:05 <danja> I think it's one of them, yep

17:09:42 <danja> ok, anyone familiar with XPath and one of them SQL-like languages

17:10:08 <danja> could you please have a look at the example FOAF data in comments at http://www.intertwingly.net/blog/1614.html

17:10:28 <danja> and come up with a good query on it...did I say please?

17:22:27 <D[a]vey> Hmm, I wonder how possible it is to work out what documents are related to what documents from the RDF data that is used within them....

17:30:09 <bitsko> that should be easy, no? common resource URIs?

17:30:29 <bitsko> er, subject/object uris

19:23:12 <eaon> eaon is now known as leofindeisen

19:25:51 <leofindeisen> leofindeisen is now known as eaon

20:10:38 <D[a]vey> D[a]vey is now known as Davey

20:49:59 <md-afk> md-afk is now known as mdupont

21:11:54 * DanC_jam reviews timbl's changes to scribe-bot/rdfn3chat.py

21:14:33 <DanC_jam> hmm... what's this for?

21:14:37 <DanC_jam> - self._kb = llyn.RDFStore()

21:14:37 <DanC_jam> + self._store = llyn.RDFStore()

21:14:37 <DanC_jam>

21:22:27 <JimJibber> JimJibber is now known as JibberJim

21:43:10 <mdupont> mdupont is now known as md-afk

21:53:13 * DanC_jam commits updates to rdfn3chat.py and README.cvs

22:24:44 <sbp> hi there, Morbus

22:38:34 <Davey> Davey is now known as D[a]vey

23:02:07 <esigler> esigler is now known as esigler_away