HCLSIG/SWANSIOC/Meetings/2009-04-17 Conference Call
Conference Details
- Date of Call: Friday April 17, 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: Susie
Agenda
- Task Reports (brief)
- Sci Discourse integration planning document Media:HCLSIG$$SWANSIOC$$Meetings$$2009-04-17_Conference_Call$IntegratingWithSciDiscourse.doc Media:HCLSIG$$SWANSIOC$$Meetings$$2009-04-17_Conference_Call$IntegratingWithSciDiscourse.pdf
- Terminology, BioRDF Task Outputs
- Roadmap discussion outline for F2F
- F2F agenda, main topics
Minutes
Attendees: Matthias, Tim, Susie, Kei, David, Scott, EricP
<Susie> topic: chart out F2F breakout
<Susie> Tim: plot roadmap for aligning with output of terminology and BioRDF
<Susie> Tim: I've drafted a document that describes the integration
<Tim> http://esw.w3.org/topic/HCLSIG/SWANSIOC/Meetings/2009-04-17_Conference_Call
<Susie> Tim: Pleased if people come up with missing pieces
<Susie> Tim: walks us through the document
<Susie> Tim: Integrating Scientific Discourse ontologies with BioRDF and Terminology
<Susie> Tim: Understand landscape, opportunities, and challenges
<Susie> Tim: Outlined phrases representing use cases
<Susie> Tim: A couple of the issues
<Susie> Tim: Ontologies versus terminologies
<Susie> Tim: There are methodology implications of that
<Susie> Tim: When would you want to use things from very controlled
<Susie> Tim: How can you provide a scaffold
<Susie> Tim: How can people work with legacy terminology
<Susie> Tim: Legacy for us includes mesh and snomed
<Susie> Eric: Will Peter Hendler be at the F2F? He's done lots of stuff in OWL
<Susie> Eric: I'd be interested in hearing his views
<Susie> Kei: SKOS isn't OWL based
<Susie> Kei: Boundaries between representations is blurred for many people
<Susie> Kei: Paolo's paper addresses some of these issues
<Susie> EricP: To me it seems like we need use cases
<ericP> Susie: components of your proposal overlap Pharma Ontology
<ericP> ... pharmaOntology is working on a high-level ontology to bridge data sources which are required for translational medicine
<ericP> ... some folks would call it a scaffold or template ontology
<ericP> ... want to use it to enable domain-specific ontologies to be bridged together
<ericP> ... work today involves id'ing 15 common roles in trans med, e.g. primary care phys, chemists, payers...
<ericP> ... id'd key questions these players would ask, as well as what parts of an ont they would depend on
<ericP> ... madden has done word-freq analysis to see which words come up in the roles
<ericP> ... at f2f, expect to look for core entities to go in the core ont
<ericP> ... to restrain size, expect to build something:
<ericP> ... .. write up 3 use cases which involve more than one role
<ericP> ... .. then build out the ont (to meet the use case)
<ericP> ... LODD folks are looking for well-designed onts to map to RDF
<ericP> ... LODD folks are thus customers of pharma ont
<ericP> ... if our template doesn't meet needs of scientific discourse, we've failed
<ericP> ... you'd be a use case for pharma ont
<ericP> ... have a poster at BioMed Ont conference
<ericP> mscottm: work freq analysis sounds like language models: a histogram of words associated with terms
<ericP> RDF gives you "and" for free
<ericP> it's innate to the data model
<Susie> word frequencies http://esw.w3.org/topic/HCLSIG/PharmaOntology/Ontology
<ericP> Susie: did nothing super-clever; just looked for most common words in the pages describing the tasks
<ericP> ... "and" at the top, but see page for relevent top words
<Susie> roles and questions http://esw.w3.org/topic/HCLSIG/PharmaOntology/Roles
<Susie> ericP: Alignment between SciDiscourse and PharmaOntology requires that alignment between terms
<Susie> scott: need to characterize what's in the named graph
<Susie> scott: trying to do coupling between resources
<kei> void link: http://users.ox.ac.uk/~zool0770/presentations/HCLS-BioRDF-Feb-09.pdf
<Susie> Tim: one thing that tends to be missing from bridging ontologies is actions
<Susie> Tim: integration is a practical question
<mscottm> http://rdfs.org/ns/void-guide
<Susie> Tim: integration occurs through the activities in companies
<mscottm> http://rdfs.org/ns/void/html
<Susie> Tim: For example, an experiment is an activity in science
<Susie> Tim: Still don't have an entity that describes what an experiment is
<Susie> Tim: I think this is a problem
<Susie> Tim: SWAN maps into more classic ontology by using 'research statement' connects to a 'reagent', 'entity', etc.
<Susie> Tim: But that isn't really enough
<Susie> Tim: Want to say what the results of work were
<Susie> David: Explain work Manchester and Southampton are doing
<Susie> David: myExperiment and BioCatalogue are similar for example
<Susie> David: Started an architecture overview board
<Susie> David: Have a project focused on Research Objects
<Susie> David: Goal to share objects across repositories
<Susie> David: Results in a aggregation
<Susie> David: Based on ORE
<Susie> Tim: Are you familiar with 'Life Sciences Entity' in SWAN?
<Susie> David: No I haven't.
<Susie> David - ORE - Object Reuse and Exchange
<ericP> Susie: lilly is also using information objects
<ericP> ... working with manchester
<Susie> Tim: Sounds like it deserves its own piece in this structure
<Susie> Scott: Thought I heard a use case relating to defining experiments
<Susie> Scott: And machine processable results
<Susie> EricP: Tim, are you talking about sentences?
<david_newman> http://openarchives.org/ore
<Susie> Tim: We are trying to create ability to have structured representations and sentences
<Susie> EricP: By sentence I meant a triple
<Susie> EricP: Do you mean triples or single key words?
<Susie> Tim: Hypotheses, research questions, and claims have to be lightweight and very flexible
<Susie> Tim: Too many things that you want to represent
<david_newman> http://rdf.myexperiment.org/example/ExperimentResourceMap <- example ore resource map for an experiment in myExperiment (this currently only includes enactment of workflows, i.e. jobs but could be extended)
<Susie> Tim: Want English sentences that are related to formal data
<Susie> Tim: Document in pharma company about a research plan
<Susie> Tim: Want to relate things to the document for improved precision and recall
<Susie> Tim: Controlled terms could just be terms, or terms in a well defined ontology
<david_newman> http://rdf.myexperiment.org/example/PackResourceMap <- example of a pack which has a more heterogenous content, ultimately a Research Object would be a more structured version of this
<Susie> EricP: Was wondering if use of a few predicates would help
<Susie> Tim: Would need examples, and work it through
<Susie> Tim: If very structured experiment (e.g. microarray) then question whether could prescribe for an experiment in general, but open to discussion
<Susie> Tim: Might want to start with the statistical approach for designing the experiment
<Susie> Tim: Nice to say upfront with predicates what you are planning to do for the analysis
<Susie> Tim: Often people do the experiment and then change the hypothesis
<Susie> David: Not working at the level of a research statement in an object
<Susie> EricP: Thinking of doing something lightweight
<Susie> David: Would have though part of defining a research object, would be to define more than rich keywords, so maybe also relationships
<Susie> EricP: Relationships could help you to identify a subset of things of interest, rather than everything about an entity
<Susie> Tim: Would be interesting to discuss our approaches at the F2F
<Susie> Tim: Should we add it to the roadmap agenda
<Susie> David: Dave de Roure will be at the F2F
<Susie> David: Would you be prepared to take the document, and talk about integrating with ORE
<Susie> David: Thinking about how best to describe ORE other than what's on the web site
<Susie> David: Some content in paper we submitted - but don't know if it has been accepted yet or not
<Susie> David: Maybe can share a copy with a handful of people
<Susie> Tim: Do people think it makes sense to discuss the Southampton and Manchester approach at the F2F
<Susie> EricP: Yep
<Susie> Tim: Need some action points
<Susie> David: I can add a couple of bullet points to Tim's document
<Susie> Tim: Could you take document and do a critical re-write of the BioRDF section
<Susie> Kei: Sure I can give it a go - will provide feedback early next week
<Susie> EricP: Is this for general consumption
<Susie> Tim: Sure, it's for everyone to provide feedback on
<Susie> Tim: Will ask Scott to provide content
<Susie> Tim: Will you write some text on alignment with Pharma Ontology
<Susie> Susie: Sure
<Susie> Tim: It'd be excellent if we could get feedback by the end of next week
<Susie> Tim: Then ask for feedback from everyone after that
<Susie> Tim: Matthias it'd be great if you could also contribute
<Susie> Tim: The SWAN IG Note is about half done, so it's getting there
<Susie> Tim: Thanks for being on the call
<Susie> Tim: Looking forward to seeing everyone at the F2F
<Susie> Tim: Next telcon on May 8