libby_: testing to show max the chump
gk: Summary of some provenance issues in RDF
gk: Following yesterday's discussion, I discovered this while trying to track down some of the EARL material. Enjoy!
bijan: The current (unreleased) SemantiChimp uses the reification model, with the reified statement being the content of the annotation, rather than the annotation itself, as it seems the Medcertain project does.
bijan: This may be unusually wieldy for SemantiChimp, given the granularity of the annotations.
gk: Following yesterday's discussion, I discovered this while trying to track down some of the EARL material. Enjoy!
bijan: The current (unreleased) SemantiChimp uses the reification model, with the reified statement being the content of the annotation, rather than the annotation itself, as it seems the Medcertain project does.
bijan: This may be unusually wieldy for SemantiChimp, given the granularity of the annotations.
chaals: Well, we get properties, but it needs some more code to make ontologies out of them...
gk: Jos asks about returning one or more answers...
gk: Possible problems of incomplete responses?
gk: GK (and ANO?) thinks a model with lazy evaluation could be interesting
gk: MikeD comments that layers of query language should(?) match layers of RDF language (e.g. basic graph, schema, OWL, etc... expect a performance drop with each level of language.
gk: EricP would like to see a common model, if not syntax, so there's a way to gateway between them.
gk: Jeremy mentioned Adobe's XMP is another -- http://www.adobe.com/products/xmp/main.html
gk: Possible problems of incomplete responses?
gk: GK (and ANO?) thinks a model with lazy evaluation could be interesting
gk: MikeD comments that layers of query language should(?) match layers of RDF language (e.g. basic graph, schema, OWL, etc... expect a performance drop with each level of language.
gk: EricP would like to see a common model, if not syntax, so there's a way to gateway between them.
gk: Jeremy mentioned Adobe's XMP is another -- http://www.adobe.com/products/xmp/main.html
libby: re QinetiQ, there's a PDF from SWWS by David Alsopp somewhere
libby: david allsopp's pdf
libby: also Sean palmer's QL
libby: some RDF query BOF notes from SWWS last year
libby: a note about jonathan robie's use of XQuery for querying RDF. the paper he did is here
libby: I just found the versa RDF QL by Uchi Ogbuji
gk: Peter Patel-Schneider did a formal semantics for XML based on the Xquery model, which was capable of subsuming RDF semantics fairly completely. I've lost the URI for it.
libby: also theres a simple query language used in RDFGateway - very similar to Guha's. This is used over a clever, rules-based database.
libby: david allsopp's pdf
libby: also Sean palmer's QL
libby: some RDF query BOF notes from SWWS last year
libby: a note about jonathan robie's use of XQuery for querying RDF. the paper he did is here
libby: I just found the versa RDF QL by Uchi Ogbuji
gk: Peter Patel-Schneider did a formal semantics for XML based on the Xquery model, which was capable of subsuming RDF semantics fairly completely. I've lost the URI for it.
libby: also theres a simple query language used in RDFGateway - very similar to Guha's. This is used over a clever, rules-based database.
gk: Currently, I don't know of a web site for the iTrust project, which is still spinning up.
libby: by damian steer; danbri had a hand in it...
danbri: Nearby, the draft SWIPE spec, basically a rough RDFization of the parameters needed by Sherlock. Since Mozilla also has a Sherlock screenscraper built-in, I wanted to try something that both damian's Sherch and Mozilla could consume.
danbri: Nearby, the draft SWIPE spec, basically a rough RDFization of the parameters needed by Sherlock. Since Mozilla also has a Sherlock screenscraper built-in, I wanted to try something that both damian's Sherch and Mozilla could consume.
JosD: This page describes a set of tools and techniques for building DAML gateways (or "front ends") to existing HTML servers.
JosD: Such gateways can be used until the site itself produces DAML directly.
JosD: there is also a very interesting CWM approach to that (already chumped)
JosD: impressive piece of SW stuff from Mike Dean/Dan Connolly
danbri: What Mike's calling a gateway, I tried calling an (rdf) Description Service; a generalisation of the pics thing...
danbri: They're called web services now :)
JosD: Such gateways can be used until the site itself produces DAML directly.
JosD: there is also a very interesting CWM approach to that (already chumped)
JosD: impressive piece of SW stuff from Mike Dean/Dan Connolly
danbri: What Mike's calling a gateway, I tried calling an (rdf) Description Service; a generalisation of the pics thing...
danbri: They're called web services now :)
danbri: Also worth a read, the old MCF white paper. These ideas cropping up with increasing regularity on RDF lists; some have even been implemented now :) :)
danbri: So how do we catalogue video co-depictions?
RalphS: 'SemaFor' - from 'Semantic Forms'
RalphS: seeAlso James Tauber's Redfoot work
em-fr: seeAlso Redfoot
RalphS: seeAlso James Tauber's Redfoot work
em-fr: seeAlso Redfoot
danbri: "(Co-depiction) turned out to be quite fun. Cataloguing digital images is expensive and time consuming, but when we focus on co-depiction metadata we create a family ofdocuments that become more interesting whenever a new document is added to the Web. And because we're using hypertext seeAlso links between RDF documents, the co-depiction data can be scattered around the Web, and harvested when required by indexing tools."
danbri: If anyone wants to collaborate on an IRC (and instant messenger) image cataloguing tool for RDFWeb/co-depiction, I'd be psyched to collaborate... (maybe a chump variant? though it'd need lots of custom hacking...)
danbri: Nearby in the Web Edd's article on IRC bots, which I just stumbled across.
edd: Not quite as sophisticated as you want, but the chumpographica is a photo collection driven by a normal chump bot. Items with <keywords>photo:...</keywords> are snarfed into the blog
edd: (you can do that on this chump by saying to the bot, e.g. X:->photos:Cannes, or invent some fancier metadata scheme, presuming you ran a copy of my software over the XML...)
libby: so all we need to add is the depiction stuff, which pretty small
AaronSw: Edd has more comments on photo/codepiction chumping
libby: re giving people identifiers, and what to do with non people, see rocky the rock rabbit
libby: a v good point Edd, same goes for JFK, other people with no email
libby: typing nodes is ok, wordnet but need common identifiers for indivoiduals
danbri: That's the fun thing about the foaf/rdfweb dataset... you can do useful(ish) things now, but it always throws up research questions or challenges for implementors
danbri: "the person that is the foaf:topic of blah" might be one approach..., where blah is a page about them.
edd: it's tough when ease-of-input is more or less your primary requirement
libby: I'm thinking now that I might try irc and squish queries over soap again for cataloguing: you need to be able to look for email addresses you know about already etc etc; seems like irc with a very clever UI might do that, and make queries where necessary. could still base it on the chump. I've tried this before with moderate success.
danbri: If anyone wants to collaborate on an IRC (and instant messenger) image cataloguing tool for RDFWeb/co-depiction, I'd be psyched to collaborate... (maybe a chump variant? though it'd need lots of custom hacking...)
danbri: Nearby in the Web Edd's article on IRC bots, which I just stumbled across.
edd: Not quite as sophisticated as you want, but the chumpographica is a photo collection driven by a normal chump bot. Items with <keywords>photo:...</keywords> are snarfed into the blog
edd: (you can do that on this chump by saying to the bot, e.g. X:->photos:Cannes, or invent some fancier metadata scheme, presuming you ran a copy of my software over the XML...)
libby: so all we need to add is the depiction stuff, which pretty small
AaronSw: Edd has more comments on photo/codepiction chumping
libby: re giving people identifiers, and what to do with non people, see rocky the rock rabbit
libby: a v good point Edd, same goes for JFK, other people with no email
libby: typing nodes is ok, wordnet but need common identifiers for indivoiduals
danbri: That's the fun thing about the foaf/rdfweb dataset... you can do useful(ish) things now, but it always throws up research questions or challenges for implementors
danbri: "the person that is the foaf:topic of blah" might be one approach..., where blah is a page about them.
edd: it's tough when ease-of-input is more or less your primary requirement
libby: I'm thinking now that I might try irc and squish queries over soap again for cataloguing: you need to be able to look for email addresses you know about already etc etc; seems like irc with a very clever UI might do that, and make queries where necessary. could still base it on the chump. I've tried this before with moderate success.
RalphS: by Mike Dean
libby: showing a cool, GraphViz based display tool
libby: showing a cool, GraphViz based display tool
RalphS: Jos de Roos
danbri: Euler, alongside Cwm, passes the RDF blank node folding problem scenario I wrote up last year.
JosD: further thaughts Design of an inference engine for the semantic web
danbri: Euler, alongside Cwm, passes the RDF blank node folding problem scenario I wrote up last year.
JosD: further thaughts Design of an inference engine for the semantic web
chaals: stores pairs - a URI and EARL assertions about that URI
RalphS: by Nadia Heninger
chaals: now stores results in annotea, so they can be extracted by Jibberjims tool along with other earl
nadia: welcoming page at http://hkn.berkeley.edu/~nadiah/tester/ is a better way to get at script because it'll be moved
chaals: now stores results in annotea, so they can be extracted by Jibberjims tool along with other earl
nadia: welcoming page at http://hkn.berkeley.edu/~nadiah/tester/ is a better way to get at script because it'll be moved
RalphS: by Nick Kew
RalphS: Dan Connolly, W3C
chaals: Not necessarily accurate help it seems :(
edd: how it was inaccurate? anyway, the latest manual is online, and is in sync with the latest source code release.
edd: how it was inaccurate? anyway, the latest manual is online, and is in sync with the latest source code release.
RalphS: Dave Beckett demoing
DanC: dude, wicked!
em-fr: a simple aggregation service for rss 1.0 w3c feeds
DanC: built with Jena
em-fr: a simple aggregation service for rss 1.0 w3c feeds
DanC: built with Jena
DanC: presented by Eric Miller, W3C
DanC: "you guys are coding all the time. I love it. I don't see how anybody could keep up with it. Except maybe Aaaron."
DanC: "you guys are coding all the time. I love it. I don't see how anybody could keep up with it. Except maybe Aaaron."
danbri: I've just committed some new stuff for database and query apps. The RDF Graph object has a toSQL() method for exporting as a PostgreSQL dump script; there's also a new library squish/squish.rb with a Squish query parser and Squish2SQL support based on Libby's original Java.
danbri: See the database and query support documentation for details, including quick 'howto' notes.
danbri: No proper tests yet, but it uses some example data from RDFWeb (Co-Depiction and FOAF-Corp/theyrule), neither of which have been properly announced yet.
danbri: See the database and query support documentation for details, including quick 'howto' notes.
danbri: No proper tests yet, but it uses some example data from RDFWeb (Co-Depiction and FOAF-Corp/theyrule), neither of which have been properly announced yet.
sbp: A new DesignIssue from TimBL, 2002-02-28, discussing how XML documents of mixed namespaces are to be interpreted
sbp: Introduces the concept of "XML functions": specific bits of XML that change the content of XML according to well-known rules. e.g. XSLT
sbp: This generally sticks to the "the document's root element namespace is the primary method for gaguing the meaning of the document" argument
sbp: and argues for a similarly top-down XML processing model, since as TimBL puts it, """XML is a tree. It is appropriate for the interpretation of the tree to be defined from the top down."""
sbp: Introduces the concept of "XML functions": specific bits of XML that change the content of XML according to well-known rules. e.g. XSLT
sbp: This generally sticks to the "the document's root element namespace is the primary method for gaguing the meaning of the document" argument
sbp: and argues for a similarly top-down XML processing model, since as TimBL puts it, """XML is a tree. It is appropriate for the interpretation of the tree to be defined from the top down."""
jah-home: developed by one of the DARPA DAML contractors
jah-home: Mark Burstein of BBN, this is a DAML
jah-home: mode for emacs - works w/Xemacs and others
jah-home: extends W3C HTML mode for DAML/RDF
jah-home: Mark Burstein of BBN, this is a DAML
jah-home: mode for emacs - works w/Xemacs and others
jah-home: extends W3C HTML mode for DAML/RDF
danbri: I tidied up the documentation and packaging a little, before working on a rewrite in Ruby.
danbri: This works against PostgreSQL triple stores of a certain layout.
danbri: So how do we catalogue video co-depictions?
danbri: This works against PostgreSQL triple stores of a certain layout.
danbri: So how do we catalogue video co-depictions?