See also: IRC log
<ericP> scribenick: ericP
what's wrong with showing a little initiative?
<mattCorkum> morning...
<david_newman> ericP: I may appear to be on a different number today as I am in a different location
<mattCorkum> morning Joanne...
<HelenaDeus> I am on the phone, but my nr should something like 713563....
<mattCorkum> are there slides that can be shared here to see what David is talking about right now?
hi folks on the bridge, i'll have slides for you in a moment
<mattCorkum> ty
<Jun> thanks
[slide 8: my experiment]
<Jun> thx eric!
[slide 10: myExperiement Features]
quite welcome
[slide 11: (workflow diagram)
[slide 12: Control over sharing]
[slide 13: A Pack]
[slide 15]
[slide 16: (Workflows screen shot)]
[slide 18: (Workflows screen shot for Adam Belloum)]
[slide 21: Google Gadgets]
[slide 23: Facebook]
[slide 24: HPC/NA]
<mattCorkum> thanks Eric...this helps to track David..
[slide 25 (lots of numbers + two buids)]
<scribe> scribenick: ericP
[slide 26: Analysis]
[video]
not uploaded, sorry
(well, that i know of)
<marco> is there a link for the video?
marco, i don't have a link
perhaps i can get it to you later
grab some coffee and delete spam for a few minutes
[slide 28]
dave deroure: we count as a success that this image does not have MyExperiment in the middle
[slide 29: For Developers]
(everal builds)
[slide 29 build: (XML, facebook, iGoogle...) ]
[slide 29 build: (RDF Store) ]
<Jun> are there different content in the RDF store and the mySQL store?
dave deroure: the SPARQL endpoint promoted a migration from API to SPARQL
there was a script migrating data between the box and the RDF Store
... now much more intimate
<Jun> thanks eric. :)
<Jun> now has a clean myExperiment ontology
<mattCorkum> where does this exist? Is there a URI to look at?
TimC: we started with a triple store and it was an impediment to agile development
[slide 30: SPARQL endpoint]
<Jun> http://rdf.myexperiment.org/ontology#
<mattCorkum> ty Jun
mattCorkum, there's link from the f2f page's myexperiment icon
<Jun> it's in page 30
[slide 32: Exporting packs]
[slide 33: The Provenance of Electoronic Data]
[slide 34: Phase 2]
[slide 35: Reuse and Symbiosis]
[slide 36: Software Design...]
[slide 37: Six Principles...]
[slide 39: e-Laboratory lifecycle]
[slide 41: e-Labs]
[slide 42: e-Labs + Research Objects]
[slide 44: Assembling e-Laboratories]
<Jun> I am going to let David know that his name is presented here, and he regreted for not being able to be here.
[slide 46]
[slide 48: onto swan-sioc diagram]
[slide 49: Characcteristics of a Research Object]
[slide 50: Thoughts]
Julia (AZ): do you see a way to help users navigate
with more than ontologies, maybe with ebusiness rules
dave deroure: yes, that's exactly what we do
we have a cookbook which we follow
carol: obesity egroup, has social-survey data linked to clinical data
... and it has policies makers doing stats over it
... we control the parms to make sure it meets the interpretation sense of the survey data
... spreadsheets, sysmo uses email and spreadsheets
... so adding spreadsheet support
ericP: do you end of writing tons of XSTL?
carol: (re taverna) yes, tried a cool systems from steve harris which described the RDF mapping, but couldn't hand to users
mscottm: would like to impose my own semantic types on the inputs and outputs
carol: jacob from pl is coming to work on this
francois: re: ericP's suggestion, i hate writing xslt, but have imposed that my taverna workflows return RDF
... encourage web services to provide RDF results
david booth: i written related stuff, google "RDF SOA"
<dbooth> Google rdf and soa: http://www.google.com/search?hl=en&q=rdf+soa
becky: have worked with large co's who usually use visio
johnM: interested in the user's desire for privacy
... in the medical realm, we have concearns about both privacy *and* non-repudiation
carol: manchester serial killer (over 300) was caught by non-repudiation mechanisms
<scribe> scribenick: Susie
Scott presents some strategy slides
Scott highlights work undertaken to date
Recaps technology trends from yesterday
Requests use cases relating to translational medicine
Link data from molecular biology to compounds
<ericP> apologies for temporarily breaking the F2F page
Harold Solbrig, Mayo - bridge between rapidly moving web technologies and health care
This is a service we don't find elsewhere
Show how tools can be used for interesting solutions
<ericP> scribenick: ericP
Olivier Bodenrieder (NLM): i think xlational med is the ultimate use case
... quoting butte: "yes we can" meaning that is bioinformatisits, we can have to tools to make the discoveries ourselves
@@@1: am cuirous about integrating @@@2 systems
... working with NCI and NCBI, competing with tim
... would be good to ineract with other systems
<scribe> scribenick: Susie
Vassil, Ontotext
Focus on use cases that are difficult to implement with existing technology
This typically requires more complex use cases
<ericP> scribenick: ericP
François Belleau (Bio2RDF): seen new SPARQL endpoints this year
... we should try to connect those things
... we don't need two versions of uniprot dumps
... need to do the dirty job of converting to RDF and agree on the correct source
... we'll be in a position to encourage the provider to adopt the RDF service
<scribe> scribenick: Susie
Oliver R, Virtual Cell Team, UCT Health Center
Hear 2 ends of spectrum
Take lots of data and measure success by number of triples
Other is the grand unification of ontologies
Like to see something in between
Take narrow domain and solve it well
For example, pathways
Developed BioPAX level 2 a while back and it's heavily used
Weaknesses are documented, but not fixed
It's a cultural/social issue, not technical issue
<marco> q
<ericP> scribenick: ericP
Elgar Pichler (AstraZeneca): would like a recipe book of how to build models and use tools
... there are several components. how do we move between these components?
... e.g. how do i convert datasource X to RDF?
<scribe> scribenick: Susie
Bosse Andersson (AZ) wants examples from Elgar's cookbook
<Jun> we talked about the same thing about 'cookbook' in the bof of linkeddata at www2009
translational medicine is a good area to focus on
don't think we should focus on too narrow an area
networking is important
<ericP> scribenick: ericP
Erick from IO Informatics: we've done some use cases with merging dbs
... when you get to medical data, need include access rights
... has been mostly under the carpet because everyone wants to connect everything
<scribe> scribenick: Susie
Zak Rhodes (IO INformatics) demonstrate value of what SemWeb can do
<ericP> scribenick: ericP
michel Dumontier: impressed by this community of diverse experts
... am generally bored by other meetings. interesetd here because of folks' experience
... we could produce [Elgar's] cookbooks
... could show how to triplify, and even find the resources
... could make a community which would allow folks to sign up for particular [triplification] and others to find them
... cookbooks need to include provenance, but also data description for discovery
... how do we get tools onto the web
... need publication to be a box in a workflow
<scribe> scribenick: Susie
Carole Goble (U Manchester) - builds software systems for LS
Agrees with Michel
Often hampered by nearly ready software, which isn't ready enough
Want key services that would make big difference to users
Not rocket science
Typiclly little things can make the difference
What are the 5 things we need as infrastructure, service or technics
really need success stories to convince people
has anyone made a scientific discovery yet using this?
This is really needed
Harness our knowledge
Ontly need one success story in order to make a difference
<ericP> scribenick: ericP
Sefano Bakoni: focus on use case, produce cookbooks, but feedback the acquired experience
... allow folks to do a cost/benifit analysis
... bring it up to the modeling level and the standardization level
<scribe> scribenick: Susie
Tim Clark (Harvard and MGH) - what Carole said!
Need to transition from hand waving as to why this is of benefit to scientists
How many people have enabled a scientist to make a discovery
'The simplest thing that can possibly work'
'Help scientists'
'What actually works'
<ericP> scribenick: ericP
Carl Taswell: spoke yesterday about portal doors
... in response to timC, i'm a scientist and my motivation is for myself
... i think this group should be an intermediary between the [general] hcls world and the individuals
... is the hcls community happy with the latest revision of e.g. owl, sparql
... the hclsig should provide a route for feedback
<scribe> scribenick: Susie
Trish (NCBO) - 1 of 7 funded biomedical centers
Some overlap with HCLS
Foster more communication between the groups
<ericP> scribenick: ericP
tim roxford (face forward): lurker here
... folks here seem to believe that semantics help get to the right abstration level
... would lay out a landscape of the stakeholder
... you typically have to drill deeply into use cases
... hard to find which ones will give you traction
... the map will help you find the low-hanging fruit
<scribe> scribenick: Susie
Becky Rhone (Independent Consultant) - external folks aren't getting what we're producing
Should I create my own RDF, or re-use RDF
Was focused on use case
But laying out ecosystem could be very import
Need 3 tiers of use cases
Techie, to middle layer, to blue sky
<ericP> scribenick: ericP
Dave How (NCI chemical informatics): i work on the core grid
... to allow scientists to share data and discovery
... we are seeing folks populate web services with good data
... but need a way scientists to answer questions like the pharma ontology questions, so easily that it's even intuitive
... that would be a compelling case for us to wrap model driven services with SPARQL endpoints
<scribe> scribenick: Susie
Helen Chen (AGFA) - minority as I'm in the trench - very scientifically focused
We have lots of data
Always want use case and prototype
Involvement for 10 years
Sold idea to a production group
EricP made that piece of software
Need industrial strength tools
Need one success story
Mapping is hard and manual
How do you visualize and validate mapping
Want to push for this for the remaining 2 years
<ericP> scribenick: ericP
Julia Kozlovsky (AstraZeneca): we should share our stuff
scribe: could put all our stuff in MyExperiment
... if we do have workflows, we can formalize them
... i think health care providers and academics are solving different problems
... would like to see some coallescing
... would like specific tools for influencing pharma
<scribe> scribenick: Susie
Paolo C (Harvard, MGH) - cookbooks are very important
Lots of tacit knowledge in HCLS
Hope integration between task forces will proceed better in the future
<ericP> scribenick: ericP
JohnMadden (Duke): agree with folks who wnat stuff now, but...
... in semweb for clinical medicine, we need privacy, security, accountability
... privacy currently silo'd in relDBs
... how you get unstructured or semistructured data into RDF (without curation)
<scribe> scribenick: Susie
Joanne Luciano (Mitre) - excited by what I see, but also frustrated
Haven't got as far as I had hoped
Like to see a visual and tool to assess what I can and can't do with existing tools
Easily see when we can easily exploit SemWeb
<ericP> scribenick: ericP
Marco: 2nd carol, timC and Becky's comment (as i'm a user)
... move beyond looking for ultimte demonstrator and instead to enablers
... there is interest in SW in bioinformatics in netherlands, but need demonstrators
... success would be when someone outside HCLS using SW
Susie Stephens (Lilly): would like to see a vibrant community of participation
... the data is a continuum, though sometimes restricting to specialties is sometimes useful
<JoanneLuciano> adding my comments: addressing John Madden's concern about in-house data / privacy issues - make accessible how to use the semweb tools with in-house clinical data (if it must stay in-house)
... real win is when a scientist uses SW and makes a discovery
... identity seems to be a sweet spot for many folks and we're not doing it at the moment
... how can an identify system manage differntn perona
ericP graph patterns for query, policy, security and discovery/federation
<JoanneLuciano> want a roadmap or a graph with bits colored in, indicating status
<scribe> scribenick: Susie
EricP (W3C) continue to use graph patterns for everything
<ericP> scribenick: ericP
dave De Roure: the comminity is motivated to do important stuff
... need more modest pieces
<JoanneLuciano> want something that tells me - if I am asking this type of question, then i need to use this technology - and includes how to use it
... they demonstrate best practice
<JoanneLuciano> must be kept simple for the user.
... what's the best example of linked data?
<scribe> scribenick: Susie
Maryann Martone (UCSD) asking people to document successes
metrics for success are also needed
Models for interaction and harmonization
Everyone has some funding
Should be cooperative competition
<JoanneLuciano> maybe when we present the usecase include links that link to how that part of the use case was solved.
Would like NIF to be involved
Educational materials are still needed
<JoanneLuciano> Users need to know what kinds of questions we can answer and can't. -- better communication
Need to communicate some things better
<marco> on Scott's invitation: enabling: e.g. cookbook ok, workflow is better (others can try and change); enable 'routine' semantisizing; enable post-analysis across domains by people outside HCLS (whatever their use case)
Gaps between communities and how they interact with digital tools
<dbooth> dbooth: 1. collaborative demos -- good vehicle for going forward, and good publicity. 2. recipes 3. privacy, security, accountability.
Coudl be a good area for a task force
Carole - success cases of adoption of semweb
<ericP> scribenick: ericP
carol: could market stuff that need not come from hcls, which demon adoption of semweb
<scribe> scribenick: Susie
Not necessarily work done by this group
Should expose these people!
<ericP> scribenick: ericP
susie: W3C has already collection of use cases and case studies
... it's not domain specific, but many are in hcls
<JoanneLuciano> I'm wondering if a use case that takes the same data and looks at it from an individual patient, a researcher, a clinician...
harold: re: anything you can do in semweb, you can do better in rdb's (johnM), what about attacking the publication angle
johnM: +1
Scott Marshall: +1 to building on what you've got
... we have stuff that work well
... would like to build incrementally
<marco> there is no link, is there?
francois: virtoso had full-text search
marco, sorry, can't read the URL from the screen
should be short - will move on to integration
<scribe> scribenick: Susie
Scott introduces some goals of task integration
Highlights desire to share and find ontologies, terminologies, assertions, workflows, experts, etc.
Oliver - pathways are a common link cross the tasks
Becky - what do you mean by workflows?
Tim - had lots of individual discussions
Interested in annotating documents
Build a framework for this
Need standardized framework as to how these things are stored
The missing piece is 'the experiment'
But lets not model it too deeply
Don't want to drill down into the level of MIAME
Experiments let you collect data
Marco - Would it be possible to create a simple, open components so I can connect a thesaurus of my choice
Use this to link to other URIs
Connect to eLab, myExperiment
Tim - Sounds very possible
Have to think about how the semantic world can add value to the web 2.0 world
It should be obvious, but lets operationalize that
<marco> or the other way around?
John - Want assertions 'that could use' ontologies or terminologies
Not an essential prerequisite
Scott - Terminology is key for integrating across tasks
<marco> I'm afraid I have to leave now. Thank you for a very good meeting, and Eric for scribing (a big help for remote people). Have a fruitful continuation!
<mscottm> Video's of Eric's lectures: http://zookma.science.uva.nl/islatv/play.php?file=20080617_other_eric_2
<mscottm> http://zookma.science.uva.nl/islatv/play.php?file=20080617_other_eric_3
<ericP> scribenick: ericP
English: ... .. ☐ is
... .. ☑ is not
... a degenerate form of first order logic
<Joanne> re: outreach... is there interest in a press release? What do we need to do in order to issue one?
marianne: i never saw this conflict. when we can use a formal term, we use it
... there is neither 0 nore 100% agreement on a term like hippocampus
... it is not desirable to let folks invent new terms whenever they want
... current challenge is to differentiate the formal an informal
... i'd like to see us figure out when you need new URIs
john madden:
<dbooth> fyi, paper about URI declarations: http://dbooth.org/2007/uri-decl/
... thinking about centralized onto markup in the wrong way
... i write a doc in e.g. .en
... i then chose to write it in a formal ont
... when you buy into an owl ont, you buy into a lot
marianne: we do need a minimal subset
john madden: when someone write something, that's their interpretation at the time
marianne: we have lexical elements like words in a dictionary
... sometimes we recognize ambiguity, and we have structures for managing it
... i don't see why this is a problem
... for instance, when we talk about a ZZZ neuron, we agree on most of the attributes of the neuron
michele: owl is nice if it flags inconsistencies to reveal our disagreements
timC: it's not just a counter example which is a refinement, it's very different hypotheses
john madden: .. 1. when you map to an ontology, you're commiting to that ontology
... .. 2. we need a way to give feedback to ontologies. do we need a formal way to track ontology inadequacies
... .. 3. sociocultural: when publishing an ontology, need pressure to publish a minimal set of mappings to other ontologies
... .. 4. RDF needs to be versionable (ontology for versioning ontologies/deprecation ?)
<dbooth> Related to versioning, l've written a bit about the URI lifecycle: http://dbooth.org/2009/lifecycle/ The write-up is still a little rough, so comments are welcome.