SV_MEETING_TITLE -- 07 Feb 2011

<mscottm> in general, it is simplest to dial directly to the MIT line if you can. It isn't as limited as the dial ins in U.K. and France.

The KEfED model https://wiki.birncommunity.org/display/NEWBIRNCC/KEfED+OWL+Model) - is a derivative from OBI - Allan Ruttenberg

Break down nanopub in several different types

Nanopub is an assertion, eg. mosquitoes cause malaria

Different types - see https://sites.google.com/site/beyondthepdf/home/program-draft/notes-from-the-breakout

<JoanneL> Hi, had trouble getting to IRC, would someone repost the link (if we're looking at slides) thanks!

<alex> same as me, where could I see the slides?

Joanne: Talking about nanopublication types- see https://sites.google.com/site/beyondthepdf/home/program-draft/notes-from-the-breakout

<davidshotton> Sorry Guys. On several attempts, both Paris and MIT numbers respond "The conference is full. No more parties can be added at this time.

How can you get citations - hedging erodes (from Burns 2011: 'Results suggest that A >B' >> 'Burns suggests that A > B" to "A>B [Burns, 2011]")

Very sorry about that David! We are victims of our own success... Need to check with W3C!

<Tim> What Anita said ! :-)

<JoanneL> David, are you able to press "operator" to ask W3C for assisitance?

<mscottm> Yes, there are 15 on the phone. That is apparently the reservation for Zakim (space for 15).

<Tim> Eric P - can you increase this limit on an ongoing basis?

Gully: looked at connections in the brain. Different bits of research - need to integrate.

<mscottm> I just asked him in another channel.

<mscottm> I'll cancel now. :)

Thnks Scott!!

<Tim> thx

<JoanneL> ABSOLUTELY - that the observations are different from the interpretation!

<Tim> Gully: observation vs. interpretation - interpretation depends (sometimes) on what observations you include / exclude

Most people focus on abstract interpretations vs. concrete observations. Need latter for metaanalysis

<Tim> This is reminiscent of Anita's ongoing points about "stories based upon data"

<JoanneL> (this is true in life too). I agree with Gully 100% that you want to stick with observations and not interpretations - so they can be compared, etc.

<Tim> interpretation being the "story part"

<Howard> There is a large literature on the theory-laden nature of observation, exp design, and interpretation; careful separating them

<Tim> Actually I think BOTH are pretty important. Look at what is published / not published - a lot depends upon the story-data relationship

I think there are 3 parts: data; observations; interpretations which correspond to 'stories that persuade (Interpretations) with data (observations and data)'

<Tim> story is very important too

<Tim> Anita, how would you distinguish data vs. observations?

Too bad Gully can't see this chat :-( will break in

Let me pull this into the phone call...

<Howard> Is "story" just another "type" of model, similar to a math model or causal model?

<Tim> We need a text to speeach robot for this call :-)

<JoanneL> Tim, agree they are both important. What I'm stressing is that the DISTINCTION must be clear. Of course interpretation is important - otherwise there is NO MEANING.

<Tim> Right, agree

<JoanneL> but meaning in different circumstances (contexts) is different.

<Tim> yup

To me data are non-textual, observations are textual representation of, er, observations, ie human view on data

Yes yes everyone reads every text differently - this is very important, take into account the mental space of the reader

So we have data - observations - interpretations - 'reader making sense of all this' = 4 levels?

<JoanneL> well, it's also important to capture the context of the data (like "consider the source") because data isn't independent of the context eitehr

<Tim> So eg the data would be an image of a Western, observation would be description of what is happening on the Western, and interpretation would be placing it in context of biology?

<JoanneL> Anita - and context too.

Ah yes that too! That's what Ed Hovy calls 'legal lifting': allowing the experimental context that a claim was made within

<Tim> the 4th level is what I would call "synthesis"

scribe: to be connected to the citation of that claim

<Tim> what you do when you read papers, ask questions, etc

Yes - which is also a type of annotation (to Tim)

<JoanneL> Yes, and this is what Gully's talking about now

<JoanneL> Capturing the context of the finding

<Tim> you mean, as in "my comments on your story>"

<Tim> ?

Yes! We'll open up the discussion to this topic now I think it'll be interesting to the group

"what you do when you read papers, ask questions, etc" = a type of annotation

<Tim> Another thing we might chat about is ramping

<Tim> separate topic

Ramping?

<Tim> Carole Goble: "on-ramps" to adoption - need to be simple

Gully: how do we break down reasoning into variables, etc.

<JoanneL> no, i mean my story in my context, then you can pick it up and put it in your context, but you know where it comes from (in terms of how I obtained it)

<Tim> thanks Joanne, I see now

Yes so two contexts: writer's/researchers context, and reader's context.

Got it.

We should write a paper about this.

<JoanneL> Tim, it's like cultural context

Gully: how do we run analysis over this model e.g. using KEfED to improve reasoning over collections to allow predictions

Experimental design such that data can be put into data repository

<JoanneL> For example, now that you're at Manchester, I can understand certain references in/from that context (I was there for a few year), and may need to "translate" to make it relevant elsewhere (this may be too general, better discussed over beer)

Nanopub: how do we make an experimental prediction based on - ugh lost train of thought!

Gully: raw data, ie single animal; averaged data; representation in a figure or table

<Howard> Sorry; there are too many tracks here that need to be separated.

hi Howard can you make a stab? I think we're talking about levels of data now - all tracks seem to lead there?

<susanna> fyi, http://mibbi.org/index.php/Main_Page

Joanne: Context within which experiment was done needs to be preserved.

<JessT> in neuroimaging data sharing we call this "meta-data"

<Howard> I spent time with philosopher Fred Suppe who has book "Semantic conceptualization of scientific theory"; "data" is theory laden

Gully: Methods nanopub: 1) Lab notebook; 2) information you need to interpret the data

<JessT> well that is part of the problem (Howard)

<Tim> it is unavoidable - from a basic neuropsych point of view, actually

<Howard> At NASA, we wanted to get satellite imagery of Earth actually *used* so we could justify paying for its maintenance; made federation gov-academic-private sector

Howard: started in neuroscience, PET scan image analysis in humans, then got into NASA Earth understanding
... Context for understanding Earth science data - three types of understanding: Government, academics, private sector
... spent some time on Semantic conceptualisation of scientific theory - observation is an act of attention; you excerpt values that you give meaning to

<JoanneL> We humans have a hard time with observation - we are so easy to interpret

<JoanneL> and to judge

<JoanneL> Agree w/ what Anita says - that people see different things (when looking at a scene)

<JessT> Or when interpreting results

<JoanneL> Radial Category - take an instance and elaborate differently

Anita: Champlain basin project: http://www.lcbp.org/ - data has very different interpretations between scientists (and different scientists, at that!), foresters/farmers, and people governing both

Joanne is a different her than she was a week ago

:-)

Howard: categories are context-based; person based

<JoanneL> I am alive, I move through time (we all do)

Panta rhei, ouden menei.

<Tim> Howard - have you read Lakoff's _Where Mathematics COmes From_?

<JoanneL> It's really cool, being human and going forward every day to experience a new day (life). A day is a little life, I think the saying goes

<Tim> JL - can't step into the same river twice?

<JoanneL> can you Tim?

<Tim> course not

πάντα χωρεῖ καὶ οὐδὲν μένει

<JoanneL> isn't tis fun!

http://en.wikipedia.org/wiki/Heraclitus#Ta_Panta_rhei.2C_.22everything_flows.22

<JessT> The question will be whether we can ever get out of the "domain-specific" context, given the theory-laden nature of language

<JessT> But it is worth a try.

<Tim> A monk asked Zhaozhou, "What is the living meaning of Zen?." Zhaozhou said, "The cypress tree in the courtyard." - Mumonkan, Case 37

Just giving a citation to Joanne's statement! From my context :-)

<JessT> :) Tim

<Tim> (to JL)

Howard: need to have a story that you're trying to make sense of; that influences how you see the dataset; how to interpret it.

<JessT> Anita re: Champlain project and dfferent groups needing different things: Faceted ontologies might help?

<JoanneL> The other ares don't know what the preconceived notions are.

<david_newman> This sounds very much like Boundary Objects - http://en.wikipedia.org/wiki/Boundary_object

<david_newman> This was part of the inspiration for research objects within myExperiment and other projects

Joanne: working with National Oceanographic Network - facilitate technology transfer to nonscientists

David - can you unmute and share thoughts?

<JoanneL> NEONINC.ORG

<david_newman> accidentally just hung up rather than unmuted

<david_newman> just calling back in

Ok!

<JoanneL> has anyone in the group heard things talked in terms of "data publishing"?

A lot of domain knowledge has to do with what a typical data finding looks like

David Shotton: what is a model, what does it look like?

DavidN are you back on?

<JoanneL> Re: David Shotton's comment about looking at a cell in an electron microsope

<Howard> That was part of the ESIP Federation work with Peter Fox and Deborah McGuinness (sp) (Joanne)

<JoanneL> Seeing vs perceiving

Joanne: difference between seeing and perceiving

DavidS: we learn what to ignore.

Hi DavidN!

<JoanneL> Learning what to ignore (someone said) --- and that's what statistics are used for

<Howard> One trouble is that none of this is "edgy"; it is an analog world very difficult to capture with balls and sticks or ER diagrams

So, wrapping up: how do we stick this into a formalism to wrap around scientific discourse?

Gully: have to start with parameters; Joanne: also the assumptions!

<Howard> I'm trying a spatial model with landmarks in space.

<Tim> or - how does one take this understanding forwardinto something useful to research scientists?

Sorry that was the project - the Ontology of Assumptions (for the Champlain model)

Yes - what Tim said.

<Tim> that is not too aboscure, opaque, byzantine or difficult to incorporate into existing practice

Howard can you talk about your work more next week?

Gully: nanopublications and domains they operate in?

<Howard> two weeks?

Howard yes please! Thank you!

Tim: DEXI model - http://esw.w3.org/HCLSIG/SWANSIOC/Actions/SWANmyExpArray
... Bestiary of nanopublications; can be simple, complex things; makes them 'sticky', help people adopt it. That is a boundary object; they are polymorphous

<Howard> let me do it in terms of Gully's notes on research objects; and maybe F0RC?

Yes wonderful; what about FoRC?

Joanne - next couple weeks pretty flat out but not to drop off;

David Newman - can you perhaps give some more data on your work on research objects and MyExperiment?

<JoanneL> I'm getting involved in 'data publishing' which is relevant and overlaps with our discussions. (the domain is earth and env sciences)

David: yes, if you could do that next week? 10 minutes would be great

David next week, Howard 2 weeks, possibly Joanne in 3 weeks (or otherwise 4)

Sorry that's DavidN next week, Howard 2 weeks, possibly JoanneL in 3 weeks (or otherwise 4)

Joanne: teaching course to turn students into researchers

<johnM> +1.31.288.aaaa is narciso

<mscottm> Hi John - sorry I'm late. BioRDF is running over.

<johnM> +1.301.827.aabb is mary

<johnM> I'd like some feedback related to our last week's discussion.

<johnM> Imagine the Description section of the report contains the

<johnM> following statement:

<johnM> "There are scattered amorphous microcalcifications in both breasts."

<johnM> So we have several ways to model this statement.

<johnM> Let's say we've created handles for theRightBreast and theLeftBreast.

<johnM> Maybe we even have decided on some handle for bothBreasts, though this is

<johnM> more difficult to decide how to do.

<johnM> Example:

<johnM> theRightBreast = [a Breast; has laterality Right; anatomicPartOf thePatient].

<johnM> (You could do this differently, say by making a restriction class).

<johnM> +20416aacc is mscottm

<johnM> I'd like some feedback related to our last week's discussion.

<johnM> Imagine the Description section of the report contains the

<johnM> following statement:

<johnM> "There are scattered amorphous microcalcifications in both breasts."

<johnM> So we have several ways to model this statement.

<johnM> Let's say we've created handles for theRightBreast and theLeftBreast.

<johnM> Maybe we even have decided on some handle for bothBreasts, though this is

<johnM> more difficult to decide how to do.

<johnM> Example:

<johnM> theRightBreast = [a Breast; has laterality Right; anatomicPartOf thePatient].

<johnM> (You could do this differently, say by making a restriction class).

<johnM> Comparison is made to exams dated: 1/1/2007 ultrasound biopsy and

<johnM> 1/1/2008 mammogram - Acme Breast Imaging.

<johnM> There are scattered fibroglandular elements in both breasts.

<johnM> There is an irregular mass with a spiculated margin in the left breast

<johnM> at 9 o'clock. This is seen in additional views. This correlates as

<johnM> palpated and with breast MRI findings.

<johnM> There also is a new 3.1 cm area of grouped fine linear and pleomorphic

<johnM> calcifications in the left breast at 1 o'clock posterior depth. These

<johnM> are seen in additional views. This correlates with breast MRI findings.

<johnM> No other suspicious masses or calcifications are seen in either breast.

<johnM> ,,,

<johnM> There are scattered fibroglandular elements in both breasts.

<johnM> *****

<johnM> bothBreast have finding [a FibroglandularElements].

<johnM> bothBreasts haveFinding [a Finding; = theFinding].

<johnM> theFinding consistsOf [a FibroglandularElement].

<johnM> rightBreast = [a Breast; anatomicPartOf thePatient].

<johnM> bothBreasts = (rightBreast, leftBreast).

<johnM> rightBreast hasFinding theFinding.

<johnM> leftBreast hasFinding theFinding.

<johnM> (rightBreast, leftBreast) hasFinding theFinding.

<johnM> rightBreast has [a FibroglandularElement].

<ericP> +1

<johnM> theFinding = {rightBreast has [a FibroglandularElement]}.

<johnM> Option 1: every single fact in the document is viewed individually as a "finding"

<johnM> Option 2: the entire document is a named graph, an inside the named graph we just state facts

<ericP> <testX> { <testX> a <PathologyText> ; :subject <breastSample7> . }

<ericP> <findingy> { <findingY> a <InterventionFlag>, <PathologyFinding> ; :resultOf <testX> }

<ericP> <findingy> { <findingY> a <PathologyFinding> ; :resultOf <testX> ; :indicator [ :location "0900"^^xsd:time ; :margin :Spiculated ] }

<johnM> {rightBreast has [a FibroglandularElement]} is a n3:falsehood.

<ericP> <findingZ> { <findingZ> a <PathologyFinding> ; :resultOf <testX> ; :indicator [ :location "0100"^^xsd:time ; :size 3.1 ; a :Calcificiation ] }

<johnM> "No microcalcifications are seen."

<johnM> We're going to have be able say this!

<johnM> thingsNotSeen = (microcalcifiations).

<mscottm> breaking up

<mscottm> ericP your sound is breaking up

<mscottm> better now

- DRAFT -

SV_MEETING_TITLE

07 Feb 2011

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output