 |
|
| FriendFeed is the easiest way to share online. Learn more » |
CSHALS 2010
FriendFeed room for the Conference on Semantics in Healthcare and Life Sciences
The CSHALS 2011 FriendFeed Group is started!
CSHALS is over for another year. I hope everybody got home OK yesterday (or today!). I am so grateful to all of our keynote speakers, session speakers, conference organizers, and attendees. Thank you!
Ted, another fantastic CSHALS conference! It was great to see how the conference has evolved over the past 3 years. More examples of practical applications of Semantics and the shift from "IF/WHEN" Semantics makes its mark on Healthcare/LifeSci to "Now we are begining to see how"" Semantics is making its mark. Great job in pulling it all together and looking forward to CSHALS 2011. Thanks, Bill
- bill hayden
Ted, as always, an excellent and mind-stimulatiung conference! As you pointed out in the closing remarks, we've come from early proof-of-concepts to production-ready and deployed applications; this is a BIG leap forward!
- Erich Gombocz
John's search is portal at: http://www.curbside.md/
I've played with Curbside.MD - very effective and concise answers returned
- Erich Gombocz
Thank you, Erich! Good to see you again at CSHALS.
- bill hayden
Can someone post a link to more info on 'Prism', the group that was discussed in the closing session this afternoon?
Direct tutorial link -- http://ibl.mdanderson.org/~jalmei... , be sure to download and install TWINKLE ... http://www.ldodds.com/project...
2 days hands workshop in Houston 2009 : http://sites.google.com/site... . This was only open to Texas Medical Center Participants and took place at Rice University. Should we try a 1/2 day version of this at CSHALS 2011 ?
Absolutely. And more along the lines. Several pure bred scientists with an itnerest in technologies and responsibilities for accommodating them want to understand what theya re about. Tutorials therefore so should be directed to the uninitiated rather than those who already know the area. Or we could have two parts/flavors. One for the uninitiated and one for those who want to be abreast of the latest in the field. Again a poll from the lunch table and other conversations.
- santha
Btw: anyone want to share a cab to Logan from the hotel around 4:45ish?
Never mind, all my flights have been cancelled - anyone care to share a ride to Vermont :-)?
- anita
Last minute ski trip, then?
- Jim McCusker
If anyone would care to join me- I'd like to drive to Vermont (burlington) welcome to co-rent a car and share the drive - all JetBlue flights that way are cancelled tonite, apparently
- anita
@anitawaard's talk: Is the problem triples, or insufficient qualification and extraction? #cshals
We'll have to see what she presents. I took her to be saying that it's the triples. I've heard this elsewhere, in particular with provenance and security. What's your take?
- Joanne Luciano
I suppose, as I am trying to say above (and in response to a question) the problem is in what people think triples represent; they are often referred to as 'facts'. Of course they are a very valuable summary, overview, faceted way to view data - but one needs to remain suspicious of the extent to which they represent 'knowledge', - was my point.
- anita
Anita, I agree, but since I have a tendency to see provenance as a way of contextualizing "facts". In a sense, I see the future of facts in SW not in triples directly, but in provenance-qualified statements that can exist as facts, opinions, observations, etc.
- Jim McCusker
Yes, I'd like to try that! Is there a representation we can play with somewhere, perhaps?
- anita
Er, no, not yet. That's work in progress. :-)
- Jim McCusker
Ah! Would love to help/know more etc.
- anita
W3C HCLS is developing a Translational Medicine Ontology: http://esw.w3.org/topic...
#CSHALS Susie Stephens of J&J speaking about W3C-HCLSIG current tasks
#cshals John Timm: Methodology for Standards-Based Biomedical and Healthcare Data Instance Generation
http://inference-web.org is the main page for the Inference Web project, related tools, and the Provenance Interlingua PML
- Deborah McGuinness
The OWL encoding of the PML can be downloaded from http://inference-web.org/wiki... (choose version 2)
- Deborah McGuinness
PML's OWL encoding can be downloaded from http://inference-web.org/wiki...
I could have used S*QL to produce my graphs for my MAGE provenance talk.
I wanted to try out S*QL when I saw it at ISWC.
provide linked data characterization according to hypothesis, evidence and relationships - this definitely will help to create trust ("truth") in assertions!
Follow-up: I believe, we all share the objective to have knowledge built from valid content in context; this can help to define the "validity" in a way which can be explored and used in ranking and weighting of text-mined assertions
- Erich Gombocz
Yes that is definitely moving in a direction I very much support and believe in! I like the concept of 'validity' - is there any formal definition of that anywhere?
- anita
Slides from my My MAGE provenance talk: http://www.slideshare.net/jpmccus... #cshals
triples may not be enough for text mining and language but they may be enough for infrastructure (repeating the argument from yesterday's discussions :-) )
Erich Gombocz and Lee Feigenbaum liked this
There is plenty of room for both triples/quads (e.g. structured knowledge) and natural language. Think of a collection of triples extracted from text using method X (human or assisted-human) as a lens into what the text is telling us. No one even really cares what the full content of a paper actually is. Everyone wants their own slice of knowledge from a paper.
- William Hayes
Yes - of course my title was polemic. I don't suggest doing away with triples - just use them intelligently and critically and where possibly augment with evidence.
- anita
Anita's talk: Why triples are not enough. Example of what is not picked up by the text processing tool. That it gets it wrong, there's no way to assess what's wrong because you only get fragments.
Yes, she's getting examples from software that tag's...but don't do a good job tagging. Triples aren't the problem, bad triples are.
- Matt Vagnoni
I hope I was able to show even good triples are not yet adequate?
- anita
What is the relationship ontology? Is it just for biology?
Eric: Is it just for biology? I was reading the relationship definitions and wasn't sure if they were designed to be general?
- Matt Vagnoni
It is for biological and biomedicine
- Eric Neumann
It is more wide than that - but has been deployed in a biological use case: see http://www.w3.org/TR...
- anita
#cshals OPM Open Provenance Model and PML Proof Markup Language
link to the MicroArray and Gene Expression Markup Language (MAGE-ML): http://www.mged.org/Workgro...
Eric Neumann liked this
good start. i think we really want a hierarchy of kindaLike. In the example yesterday we probably want a specimenKindaLike...
- Deborah McGuinness
Jim McCusker is talking about Representing Microarray Experiment Metadata Using Provenance Models.
PhenX uses the NCI Metathesaurus, http://ncimeta.nci.nih.gov/.
Josh Levy is talking about the PhenX toolkit: https://www.phenxtoolkit.org/
Other ways to read this feed:
