See also: IRC log
<trackbot> Date: 16 June 2015
<Lloyd> zakim ipcaller is lloyd
ANNOUNCEMENT: We are NOT using the w3c teleconference bridge today. We are using the webex bridge:
Here's the webex info:
Webex for teleconference:
https://mit.webex.com/mit/j.php?MTID=m5cd1bd8bb36825b9c4b369fd664bbb62
dial-up: +1-617-324-0000 Access code: 645 777 110
Meeting password: 4257
please hang up from w3c bridge and call the webex teleconference number
<Marc_Twagirumukiza> Hi, am calling in with my Laptop
Anil's slides: http://tinyurl.com/salus20150616
Anil: Background in CS. Was tech
lead in SALUS project, officially ended in APril.
... Was an EU project with 10 partners.
slide 2
slide 5
slide 9
slide 10
slide 11
CIM = Common Information Model
slide 12
slide 13
slide 15
slide 16
slide 17
slide 18
slide 19
slide 20
<ericP> Euler
http://eulersharp.sourceforge.net/
anil: Used Euler for semantic translation: http://eulersharp.sourceforge.net/
slide 23
slide 24
slide 25
slide 26
slide 27
anil: Data element exchange
profile is being prototyped.
... Will allow definitions to be shared.
... Main motivation was for form designers to auto populate
forms.
... E.g., during clinical trial, if researcher is designing a
form, it can be associated with a data element.
... Modeled as abstract data elements.
dbooth: What are the plans for SALUS now that the project has officially ended?
anil: For the whole framework, no
concrete plan to use it all. But for different components,
partners are building more tools.
... Metadata registry can also be used in other domains.
... SALUS is not planning to continue the framework, but
components
<Zakim> ericP, you wanted to ask about common information model
eric: You have two levels of
mapping: 1. map the different terminologies; 2. map the
information model.
... You ontologized the CCDA model, lifting, and worked toward
CIM.
anil: Yes, first CCD and then CIM.
eric: Was there an advantage to that over picking one of the physical models and using it?
anil: biggest advantage was to leave the mapping to the outside world.
eric: Does that mean you matierlized the data in your info model?
anil: Yes.
eric: so you wrote n mappings from the source models to your CIM?
anil: Yes.
eric: I'm guessing a lot of our future work is to develop our ability to do those mappings. Your wisdom on that?
anil: at the beginning we tried
to build a CIM to cover all data requirements based on our use
cases.
... But it's not possible to meet all field requirements. One
size does not fit all.
... And it's a fact that tthere will be many models around.
Thats' why we built this metadata registry.
... using ISO 11179.
eric: How much did you model? Patients and observations? Did you get into constellations of observations, or just single observations?
anil: Not much detail, just codes
of observations.
... On slide 6 we queried for some conditions and analyzed
common conditions that happened at the same time. We can ask
for multiple conditions.
dbooth: Data element exchange profiles: Sharable? Reusable?
anil: Most are now available
paul: slide 28 also mentiones data element exchange
anil: they allow machines to access the details of the data element definition, so that a SPARQL or other script can use that to access data.
dbooth: Assessment: success of using RDF? What benefits and challenges did you experience?
anil: Challenge was
performance.
... Executing the reasoners for inference requires much more
processing time.
... But the models and changing them are easier in RDF.
... And if data is not available it doesnn't break, it just
doesn't show it.
... Also doing the mappings for the data transformations.
dbooth: More specifics about performance challenges? what was used and what did you run into?
anil: Machine w 16 cores and
64GB, we had difficulty processing Lombardi region data, 16M
patients, 500M records in conditions, 300M prescription
records.
... Answering a query like shown takes about 2 days, mostly
doing translations: XML to RDF and also to the CIM.
... Euler is not enough on this hardward.
... We tried to implement our own map-reduce algorithm to
overcome this, and might be better to use big data
framework.
ADJOURNED
This is scribe.perl Revision: 1.140 of Date: 2014-11-06 18:16:30 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) No ScribeNick specified. Guessing ScribeNick: dbooth Inferring Scribes: dbooth WARNING: Replacing list of attendees. Old list: +33.1.73.71.aaaa DavidB [IPcaller] +1.310.266.aabb New list: +1.310.266.aaaa Default Present: +1.310.266.aaaa WARNING: Replacing previous Present list. (Old list: Kerstin, Tony, Brian_Pech, Marc, Darrell) Use 'Present+ ... ' if you meant to add people without replacing the list, such as: <dbooth> Present+ Kerstin_Forsberg, Tony_Mallia, Brian_Pech, Marc, Darrell_Woelk, Anil_Sinaci, Charlie_Mead, Lloyd_McKenzie, EricP, Paul_K._Courtney, Claude_Nanjo, David_Booth Present: Kerstin_Forsberg Tony_Mallia Brian_Pech Marc Darrell_Woelk Anil_Sinaci Charlie_Mead Lloyd_McKenzie EricP Paul_K._Courtney Claude_Nanjo David_Booth Found Date: 16 Jun 2015 Guessing minutes URL: http://www.w3.org/2015/06/16-hcls-minutes.html People with action items:[End of scribe.perl diagnostic output]