See also: IRC log
<scribe> scribenick: ericP
marianne martone
kei chung: SW, neuroscience, research
dave booth: (cleveland clinic) patient records combined
dave derour (Univ sohamption): web 3.0, workflow, bassplayer
olivier B (): ontolo di
(NCI): terminologies, biomed vocabs, software infrastructure
Harold S (Mayo): term, ont, DI
dave how (NCI): xlat research
oliver (cell modeling): pathways, ...
[missed]
helger (AZ): km frisbee
bosse (AZ):
carl taswell (): brain imaging
[missed]
paolo
michel dumontier:
joanne l (Miter): computation methods for personalize med
[missed]
ericP (W3C): biker bum astray
anita
scottM
karen (W3C)
carol gobel (manchester): shoes
stephane (); identifiers
(ontotext): ontologies
Bill Hayden
Zack Rhoades
[susie does intro]
-> http://esw.w3.org/topic/HCLSIG/Meetings/2009-04-30_F2F?action=AttachFile&do=get&target=SemanticWebStateOfTheUnion.ppt Semantic Web State of the Union: Strengths, weaknesses, opportunities, and threats
welcome adewa
mscottm, if anyone e.g. skypes you to ask how to connect to irc, i've added instructions to the bottom of http://esw.w3.org/topic/HCLSIG/Meetings/2009-04-30_F2F on how to use mibbit
<adewa> Hi David, glad to see you here!
<adewa> Can you hear the talks?
<jun> hi David:)
questions:
<scribe> scribenick: ericP
joanne: details about dillution of term "semantics") threat?
s/that/dillution of term "semantics"/
susie: if folks think cloud computing solves semweb problems, they won't persue semweb solns
michel: evolution of tech is a threat. [can't use today's stuff tomorrow]
karen: timescale of tools is a big issue
... business wants solutions today
... business not expecting to do to much geeking
carol: impressed by co's which use RDF but don't use "Semantics" keyword
... e.g. Amalga (MS) uses RDF, but doesn't advertise that
... this tech is becoming mainstream in pharmas
... we must not be holier-than-thow, e.g. appreciate only open data
anita: SW was over-evangelized prematurely
... using the term isn't helping the cause, using the tech is helping
<scribe> unknown: have to avoid terms which have been using to create silos in the past
UNKNOWN_SPEAKER: need a simple paper about how SW differs from e.g. RDBs, Excel
timC: folks have heard "we solve all your problems" too many time
... need to make this one of the tools in your bag
<mscottm> David and Jun: we're uploading the slides.. just a minute..
<jun> thanks Scott
<mscottm> Thanks Eric!
welcome
they were huge -- had to push them on to www.w3.org site
<mscottm> Maryann said that they were 25MB.. not 25T! ;)
[slide: "Google" for Neuroscience +3]
[slide 16: level 2: Disco]
[slide 25: To Map or Not to Map]
[slide 28: Concept-based search]
[slide 32: NIFSTD:...]
<David_Shotton> Eric, your notes are v helpful. Thanks!
[slide 38: Neurolex]
quite welcome David_Shotton
[slide 41: NIF in practice]
<David_Shotton> Excellent talk
DavidB: have you published your best practices?
Maryann: just about
... we're not sure if these things are really barriers
... discussions happen here (hcls), but not in neuroscience
... expect to push them out soon as "from our experience, here are five problems"
DavidB: you have not included commercial sites
Maryann: legacy from the Neuroscience gateway
... we're roughly adhering to this
... didn't want to flood the site with commercial products
... want to have ranking systems in place first
... we ask our users if they want to see commercial tools
... i believe NIF needs to both push community and be responsive
johnB: what search results do you get from NIF vs. google?
Marryann: there is a lot of spurious stuff in NIFWEB
... in the early days, we needed to seed with some results
... we do stop crawls if we find fals results
<mscottm> ...focus more on educational sites but it's meant to be a broad spectrum.
<mscottm> Elgar: How much of a headache is the synchronization process? Do you have recommendations?
<mscottm> Maryann: One problem was deprecated OBI classes. Bill Bug made a set of proxy classes and update them. But we don't have a magic solution.
<David_Shotton> Signing off - many thanks.
<jun> thanks Eric, for the notes
johnM, i think your sound will be fine if you want to do the speaking while one of us clicks your slides
you'll have to keep it as short as possible. we are, of course, behind schedule
<johnM> We can try that, sure.
<mscottm> Great! Thanks John.
@@@2: At BioIT world, Susie referred to existing drugDBs
scribe: do you have something like that for SNPs?
Kei: not at present
ericN: see also, Sean Moody, and mutedb (SP?)
[slide: Published Data Sets]
note: esw wiki has links for "...&do=get" as well as "...&do=view" . the latter does not work in links
<johnM> thx
i've fixed the links in http://esw.w3.org/topic/HCLSIG/Meetings/2009-04-30_F2F
[slide 11: Conclusions]
Anita: when the BBC publishes data, is that bits of text, numbers?
Susie: we've been looking at semi-structured data represented in XML
ericP: assume you've been focusing on data who's intended consumer is machines
Anita: so what's the BBC data/
?
Dave DeRoure: BBC data is typically cataloge info
mscottm: is typically data from RDBs?
Kei: data doesn't automatically generate knowledge. requires lots of human processing
Susie: i think linked data does work better for some granularities of data than others
... e.g. linking Berlin is better than linking the number 2
... don't want to use it for publishing whole abstracts
Anita: can use representations of scientific data sets. you don't mean as produceed by experiments?
... we have this issue at Elsevier. We can impose some structure on our documents and project it into RDF. but what uses cases does it support?
Becky: what's your coverage of diseases?
Susie: every drug that folks are developing drugs for
... per clinicaltrials.gov
Olivier: from 2001 on
Bosse: there is a lot of text info in LODD...
<dbooth> Re whether text qualifies as data, it's a qualitative judgement call.
Joanne: how difficult was it to linke these datasets?
susie: they sometimes talk about slightly different things
... e.g. active compound vs. the whole thing (with buffers)
Joanne: of your target, how much coverage did you get?
Susie: hard to quantify
... data providers would be best folks to establish the linked data
Joanne: for follow-up, what makes a good quality link?
[slide 2: Pharma Ontology]
[slide 4: Goals & Deliverables]
[slide 7: Existing Resources]
[slide 11: Summary]
[slide 3: Medical terminologies today]
<Joanne> where are the slides?
<Joanne> ignore question (i see the link now). thanks
[slide 6: Taskforce repository]
[slide 8: (screenshot)]
<dbooth> I'm amused that the value of SKOS was seen as its *lack* of formal semantics.
<anitawaard> why amused? is that not the point of skos?
it's ironic in the SW context 'cause we're always arguing to add semantics to stuff
what we rarely say, but really mean, is "lable stuff with the *appropriate* semantics"
<anitawaard> I thought he point of skos was to allow the 'broader-narrower than' relationships - because people (if not machines) really like them?
<anitawaard> good enough semantics, kid of like good-enough parenting...
<anitawaard> kid -> kind
[slide 11: Immediate next steps]
note the proliferation of the "Raw Data Now" mime
<dbooth> y, it's not quite the *lack* of semantics that bring value, but the presence of semantics that are not overly constraining.
<anitawaard> or overly incomprehensible...
<dbooth> y
yeah, rdfs:seeAlso is used quite a lot because it has almost no interpreation beyond, get this, "see also"
<anitawaard> wonderful
otoh, when ever you try to use rdfs:seeAlso, you have no idea if it was coded for the scenario in which you intend to use it
<anitawaard> we are successfully dumblng down the standards to our level
<anitawaard> :D
<dbooth> :)
[slide 13: Social Challenges]
<anitawaard> our systems are too smart for us...
<anitawaard> we have to learn what they want us to do at some point in the future -
<anitawaard> but not today...
<anitawaard> :P
<anitawaard> I like the last bullet - RDF connects us to our documents...
dbooth: did you mean URNs?
ericP: URIs is the superset
<dbooth> URI is the general term. URNs and URLs are a subset of URIs.
anitawaard, dbooth, it would be interesting to look at existing uses of rdfs:seeAlso and with an eye towards factoring common uses of it which would benefit from more explicit terms
johnM, tx a zillion!
<anitawaard> is "rdfs:/eeAlso" = rdf-a?
<johnM> David, sorry you are quite right. (I never can keep that straight!!!)
<anitawaard> or am i missing something (quite likely)
<dbooth> ericP, Pat Hayes a few months ago threatened/offered/proposed to write a new version of RDF, which would "fix" various omissions/mistakes, such as the lack of a way to name a graph (duh!) and other things.
<dbooth> If he goes forward, maybe he'd be open to putting something more in.
[slide 5: Use Case Step-Through]
-> http://www.w3.org/2009/Talks/0225-cshals-ep/ pictorial click-through of the COI query transformation
slides 1-10,23-27
[COI slide 11: Medication]
[slide 12: Criteria in SPARQL]
[slide 15: Drug Ontology by Stanford]
[slide 27]
[slide 28]
[slide 31]
[slide 32]
ericN: would like to discuss drug ontologies
[slide 4: Anser]
salut, marco
[slide 5: (chart)]
[slide 5: (diagram)]
[slide 8]
[slide 9: Scientific knowledge ecosystem]
<jun> thanks. got it:)
[slide 11: Scientific Discourse Task Status April 20 2009]
<ChristiDenney> my apologies. i have been called into a meeting and i will not be able to attend the pharma ontology breakout.
ChristiDenney, i'll relay to susie
CarlT: intested in connecting brain behavoir with imaging
<anita> PortalDoors
CarlT: an overall scheme called PortalDoors addresses most of what's been brought up here
<ChristiDenney> thank you eric! :-)
<david_newman> Hi. Is there something clever I am suppose to do if I want to join a particular breakout group from the afternoon session? i.e. the scientific discourse one. The wiki suggest there is just one general line
<mscottm> breakouts begin at 14:05
<mscottm> you need to join hcls for biordf, hcls1 for Pharma Ontology, hcls2 for Sci Discourse
<david_newman> thanks.
<mscottm> I suppose that you supply the analogous codes to Zakim.. Checking.
<david_newman> so it would be 42572 for sci discourse then?
<mscottm> yes
<david_newman> great
hello, who just joined the phone bridge?
julia: you've implemented something that is a lot like the semantic web
... why not just use semweb technology?
<marco> marco just joined by phone (skype)
carl: rules on declaring a URI is up to the registry
<david_newman> ericP: I have joined on the hcls2 phone branch
david_newman, on hcls1 or 2?
<david_newman> david_newman is +0196266aadd on hcls2
<mscottm> we're just now splitting into different rooms for breakouts
<HelenaDeus> sorry, i lost the connection. which number shoudl I call for the Bio2RDF breakout?
<junzhao> you should just dial the general number on the wiki page
<junzhao> it is used for biordf breakout at the moment. scott is talking
<HelenaDeus> cool, thanks :)
<junzhao> you are welcome:)
<HelenaDeus> is there a url for the slides being presented in each of the breakout sessions?
<mscottm> Ouch. Francois just started but I'm afraid that we don't have a copy to upload - I'll ask his colleague.
<HelenaDeus> yeah... i can hear him speak but nothing to look at :)
<mscottm> I talked to Marc-Alexandre. He says that it is on slide share and he will send it to me.
<mscottm> He is showing the Bio2RDF could map of 2,3 billion triples in 2009
<mscottm> Voila! http://www.slideshare.net/fbelleau/bio2rdf-w3c-hcls2009
<mscottm> He is on "Why do it?"
<HelenaDeus> aha! Thank you so much