HCLSIG/SWANSIOC/Meetings/2009-05-22 Conference Call

From W3C Wiki

Conference Details

  • Date of Call: Friday May 22, 2009
  • Time of Call: 11:00am Eastern Time
  • Dial-In #: +1.617.761.6200 (Cambridge, MA)
  • Dial-In #: +33.4.89.06.34.99 (Nice, France)
  • Dial-In #: +44.117.370.6152 (Bristol, UK)
  • Participant Access Code: 4257 ("HCLS")
  • IRC Channel: irc.w3.org port 6665 channel #HCLS (see W3C IRC page for details, or see Web IRC)
  • Duration: ~1 hour
  • Convener: Tim Clark
  • Scribe: Matthias Samwald

Agenda

IRC transcript

Attendees: Tim Clark, Anita de Waard, Julia_Kozlovsky, Kei Cheung, Scott Marshall, Eric P, David Newman, Jack Park



(17:07:28)  Tim: What i propose as a topic: the HypER meeting in Amsterdam
(17:08:52) mscottm [marshall@81.69.36.208] hat den Raum betreten.
(17:09:02)  .Introduction by Attendees: Anita de Waard (Elsevier) 
(17:09:29)  Julia_Kozlovsky: background in text mining, joined F2F, starting to get invovled in the group
(17:09:50) mscottm: Zakim, who is on the phone?
(17:10:27) Zakim [rrs-bridgg@128.30.52.30] hat den Raum betreten.
(17:10:29) mscottm: Zakim, who is on the phone?
(17:10:30) Zakim: sorry, mscottm, I don't know what conference this is
(17:10:30) Zakim: On IRC I see mscottm, kei, Julia_Kozlovsky, matthias_samwald, LeeF, Cloud, ericP
(17:10:47)  (further introductions by Tim, Kei)
(17:14:18)  (Introductions by Jack (from group of simon buckingham-shum), matthias)
(17:14:45)  (introduction by dave (MyExperiment), scott)
(17:15:58)  Tim: scott, a colleague of you, andrew gibson, was also at the HypER meeting
(17:16:11) david_newman [drn05r@152.78.61.24] hat den Raum betreten.
(17:16:20)  Scott: Andrew co-authored an article with me and others called "Structuring Hypothetical Knowledge"
(17:17:02)  ... factoring out "p53" as an instance of a term (not as an instance of a protein)
(17:17:40)  (introduction by EricP)
(17:18:13)  tim: the Future of Research Communication conference (FoRC) is coming up in about a year
(17:18:39)  ... ericP will also be involved in FoRC
(17:19:42)  ... have a look at the e-mail i sent out today, there is a link to a page describing intersections between HypER and the scientific discourse task.
(17:20:08)  anita: there are several partly similar initiatives
(17:20:37)  ... around the idea of representing scientific statements and links to evidence that these statements are based on
(17:21:07)  ... E.g., Cohere (simon buckingham-shum), SALT and Connex (DERI),
(17:21:30)  ... Liquid Publications (group of Fausto, University of Trento)
(17:21:47)  ... then of course SWAN (tim clark)
(17:22:08)  ... and my own work
(17:22:24)  ... i also started a collaboration on this with agnes sandor (Xerox)
(17:22:46)  ... it seemed like a good idea to compare notes. this is what we did at the HypER meeting
(17:22:55)  ... we had several presentations.
(17:23:13)  ... http://hyp-er.wik.is/
(17:23:26)  ... (this is the wiki page that will be used for documentation)
(17:24:56)  ericp: hypothesis can come from NLP, human curation, or author annotation. how where these represented at HypER?
(17:25:04)  tim: all approaches were represented
(17:25:18)  ... each of these has problems.
(17:25:29)  ... human annotation does not scale
(17:25:55)  ... NLP has either too many false positives or too little coverage
(17:26:09)  ... and author compliance in the SDA pilot was bad
(17:26:33)  ... a compromise is needed
(17:27:12)  Julia: can we come up with a cookbook? (e.g., for these scenarios, these approaches work best etc.)
(17:27:58)  tim: sometimes we just need to try to find out (e.g., SDA)
(17:28:46)  anita: SWAN has a lot of take up in its community, SDA has seen less uptake
(17:29:13)  ... communities, subject areas (e.g. neuroscience is different than protein interactions) and technologies...
(17:30:17)  tim: if the preliminary conclusion is that none of these solutions is sufficient, i guess we need to find a shared information model to reach an integration / synergy
(17:31:09)  ericp: one field of overlap i see is query federation. 
(17:31:39)  ... "i want NLP annotations with this particular confidence etc." -- that would be interesting for query  federation.
(17:32:01)  tim: http://esw.w3.org/topic/HypER_%E2%88%A9_SciDiscourse_Relevant_Tasks
(17:33:25)  ... alignment not also with SIOC, but also with SALT 
(17:34:00)  ... then we also have overlap with tagging-based approaches (aTags etc.)
(17:36:19)  matthias: sounds good, alignment makes sense and should be easy to accomplish
(17:36:41)  scott: tim, should we get paolo involved, who should get together?
(17:37:12)  tim: ideally matthias, paolo, someone from siggis group
(17:38:41)  kei: is there some simple example that we could use to illustrate those different tagging approaches?
(17:39:07)  ... not a formal presentation, but maybe a wiki page
(17:39:51)  tim: i can post some examples of SWAN
(17:39:58)  matthias: i can do so for aTags
(17:40:10)  tim: we can add them to the wiki page that i referred to
(17:41:14)  anita: are there any use-cases that we could try to deal with?
(17:42:51)  tim: matthias, pick some wiki page in SciDiscourse to place this.
(17:43:45)  ... we also can have talks at the next telecon in 2 weeks
(17:44:19) mscottm: q+
(17:44:19) ***Zakim sees mscottm on the speaker queue
(17:45:17)  ... we also will try to get someone from Siggi Handschuh's group to present
(17:45:32)  Kei: this could help to align work between task forces.
(17:46:17)  scott: matthias, kei and I already talked about storing aTags in an RDF repository
(17:47:30)  ... if we have them stored in a system and SWAN statements in another system about the same subject matter, we could try to use this for finding points of alignment.
(17:49:45)  matthias: i think we should try to avoid too much redundancy
(17:50:10)  tim: i think it is very clear how they should be aligned if you look at it.
(17:51:27)  ... we could take examples and work them through on paper
(17:55:27)  scott: i have a question about SWAT4LS, we are organizing it this year in november
(17:55:58)  ... november 20 might be a possible date. would this date work for all people on this call?
(17:57:03) mscottm: http://www.swat4ls.org
(17:57:05)  (nobody objects)