HCLSIG/SWANSIOC/Meetings/2009-09-18 Conference Call
Conference Details
- Date of Call: Friday Sept 18, 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: TBD
Agenda
- OBO Relationship Ontology - Larry Hunter
- 2009-2010 Work Plan (Contd.)
- AOB
Useful Links
- Final SWAN & SIOC IG Notes.
SIOC, SIOC Types and Health Care and Life Sciences Semantic Web Applications in Neuromedicine (SWAN) Ontology SWAN/SIOC: Alignment between the SWAN and SIOC ontologies
Minutes
Attendees: Tim, Matthias, David, Joanne, Scott, Larry, Anita, Michelle, Tudor, Cartig, Susie, Jack
<Susie> Larry Hunter Presentation
<Susie> Familiar with OBO & OBO Founder
<Susie> Interested in using community curated ontologies for NLP
<Susie> Need relationships between factual assertions, and the originating text
<Susie> Need the evidence
<Susie> Participated in a Relation Ontology piece of work
<Susie> Convened a F2F meeting in Colorado
<Susie> Talked about commitments with OBO Foundary
<Susie> They want a very minimal set
<Susie> Willing to put on the pending list, 2 relations
<Susie> 1. Supports, and isSupportedBy
<Susie> 2. Contradicts, and isContradictedBy
<Susie> Barry Smith and Chris Mungall want details on epistemic relations
<Susie> Also suggested something is going to happen in this area
<Susie> Spoke with Anita about epistemic relations and community annotation
<Susie> Want to least set down a marker to say this exists
<Susie> And try to work in harmony
<Susie> At F2F, identified all sorts of relationships
<Susie> Take away from RO meeting was to keep them out of RO ontology
<Susie> But have different ontology built on GO evidence codes
<Susie> Michelle: GO uses a subset of relations from Evidence Code ontology
<Susie> Use hierarchy of relations
<Susie> And link to epistemic codes
<Susie> Information Artifact Ontology
<Susie> Could be domain for some of these relations
<Susie> URLs pointing to the pending RO, Evidence Ontology, and ...
<Susie> Want to bring stuff back from this call to these other communities
<Susie> Also happy to answer questions
<Susie> Tim: This effort is very focused on harmonization across discourse and discourse ontologies
<Susie> Tim: Last year did harmonization between SIOC and SWAN
<Susie> Tim: Jack working with Simon Buckingham Shum, Tudor (DERI), David Neumann (myExperiment), Scott (Terminology), Anita (Elsevier), etc.
<Susie> Tim: Some active in annotating documents and linking to controlled terms in ontologies
<Susie> Tim: David Shotton is missing
<Susie> Tim: Let's open up for questions
<Susie> How do you use these relationships
<Susie> Larry: Trying to do IE
<Susie> Trying: Create KB tying genes and gene products
<Susie> Larry: Try to understand what is already known that relates to the results of experiments
<Susie> Larry: Sometimes not much work done
<Susie> Larry: Sometimes (50%) lots of work but from different field
<Susie> Larry: Postdocs need to figure it all out
<Susie> Larry: First application is the Hanalyzer
<Susie> Larry: Use NLP to build networks
<Susie> Larry: Nodes are genes
<Susie> Larry: Links are existing data
<Susie> Larry: Overlay experimental and literature networks
<Susie> Larry: See what's previously known
<Susie> Larry: Cytoscape plugin used to see more information relating to the nodes or links
<Susie> Larry: Inference is used to help build the networks\
<Susie> Larry: Linkages in knowledge networks to literature is why we're intrested in epistemic relations
<Susie> Larry: Try to be a user, rather than a builder of RO
<Susie> Larry: See bunch of pending relations that have been requested
<Susie> Larry: Interested in getting these type of relations in
<Susie> Larry: Discussed with Kathy Wu at ICBO
<Susie> Larry: She's recently moved from Georgetown to Delaware
<Susie> Larry: Kathy is very focused on the protein ontology
<Susie> Tim: You have 'supports' and 'contradicts', do they have evidence codes?
<Susie> Tim: The text is just a string, the assertion is formalized in the KB?
<Susie> Larry: Correct
<Susie> Tim: Do you use reification?
<Susie> Larry: In our code, we use Protege internally, we just have a slot for it
<Susie> Larry: Likely not going to do it in a formalization
<Susie> Larry: That's what you guys are good at
<Susie> Tim: People interested in extracting info from text
<Susie> Tim: More interested in higher level models
<mscottm> Zakim, who is making noise?
<Susie> Tim: Also interest in relationships between assertions
<Susie> Tim: Are you interested in this approach
<Susie> Larry: Grad student - Elizabeth White
<Susie> Larry: Working on thesis on more tailored model for discourse, looking at what's aligned and what contradicts
<Susie> Larry: Think our approach is more abstract, rather than yours
<Susie> Larry: Don't think text is nuanced enough
<Susie> Larry: want users to be able to analyze their own data
<Susie> Larry: We can get away with less detailed models
<Susie> Larry: Would be interested in community models
<Susie> Larry: And how we can start to populate them
<Susie> Tim: Very though for an automated approach
<Susie> Tim: Have application, with 100s of users in AD community
<Susie> Tim: Have part time curator
<Susie> Tim: People also submit content
<Susie> Tim: Usually know who you are arguing with or supporting
<Susie> Larry: SWAN is awesome
<Susie> Larry: Our focus is on breadth
<Susie> Larry: Therefore have different needs
<Susie> Larry: But are complimentary
<Susie> Tim: Have you thought about when a large lab is doing an experiment
<Susie> Tim: And want to tie them to results in that way
<Susie> Larry: Not so much
<Susie> Larry: As we really focus on the literature
<Susie> Larry: Focus a little in OBI
<Susie> Larry: Have heard concerns and difficulties that people have in tying evidence to discourse
<Susie> Larry: Capture the experimental protocol, the goal of the investigation, the outcome (type of data and transformation), and conclusions
<Susie> Anita: Interesting. Close to what I'm trying to pull together
<Susie> Larry: OBI is pretty active
<Susie> Anita: Have 2 symetrical relations
<Susie> Anita: What were the ones you didn't get?
<Susie> Larry: Original proposal on RO web site
<Susie> Larry: We had a hierarchy of relations
<Susie> Larry: The foundary wanted them as a type hierarch, and wanted to keep the relations minimal
<Susie> Larry: To us this is a format difference
<Susie> Larry: We ended up with just the 2 because of their desire to have min relations, and put everything else into evidence code
<mscottm> +1
<Susie> Larry: I also wanted a superclass for 'isRelevantTo"
<Susie> Larry: Barry wasn't keen on that as he felts that evidence could be neutral
<Susie> Larry: Not a philosopher, I just want something I can use
<Susie> Larry: So can use other approaches
<Susie> Scott: That's what was trying to mention in my mail, can include in the application
<Susie> Tim: Common for people to cite relevant evidence, but then don't commit
<Susie> Larry: That's OK.
<Susie> Larry: RO always required definitions
<Susie> Larry: couldn't come up with good definition of 'neutral'
<Susie> Larry: Would be fine with a logical combination
<Susie> Scott: Where are holes in ontologies for knowledge provenance?
<Susie> Larry: Not familiar with what you are doing
<Susie> Larry: Let's make sure everyone is compatible with 2 evidence codes in the RO
<Susie> Scott: Information Artifact Ontology and evidence codes fill some gaps
<Susie> Scott: Are we missing describing knowledge with provenance, or are there other gaps?
<Susie> Larry: Don't seem gaps, but processes
<Susie> Larry: Want RO to be compatible
<Susie> Tim: Makes a lot of sense
<Susie> Tim: Where do projects overlap
<Susie> Tim: Overlap can be large or small
<Susie> Tim: Some folks are working on annotation ontologies for publications
<Susie> Tim: We could do a side call on that
<Susie> Larry: Marking up biomedical jourmal articles
<Susie> Larry: Will participate as needed
<Susie> Tim: During last, we decided next target for harmonization included the SALT ontology, myExperiment, and CITO
<Susie> Scott: BioCatalog is tied in with myExperiment
<Susie> Scott: Let people register web services
<Susie> Scott: So are making assertions about the web services
<Susie> Scott: This is Carole Goble's work, and closely connected to myExperiment
<Susie> Scott: Should also remember Matthias' aTag
<Susie> Larry: My project is called CRAFT
<Susie> Tim: Lots of common interest
<mscottm> Colorado Richly Annotated Full Text, I think.
<Susie> Larry: The other project is Hanalyzer
<Susie> Tim: Talked about order of the integration work
<Susie> Tim: We can do things in parallel if enough people are working on it
<Susie> Scott: Getting SIOC Export for various types of discourse
<Susie> Scott: Then could SPARQL across systems using the same or similar terms to gather evidence
<Susie> Tudor: Proposing using SIOC as an upper layer ontology to bridge representations
<Susie> Scott: Would shy away from calling it an upper layer ontology
<Susie> Tudor: Won't that be limited, given SIOC doesn't capture discourse
<Susie> Tim: Where just SIOC, then that's what you get
<Susie> Tim: If it's more granular, then people can use SWAN
<Susie> Tim: We have something like this for our task list for the group
<Susie> Tim: Tudor has proposal for harmonizing SWAN, SIOC and SALT, and swanalto (sp?)
<Susie> Tudor: The 4th ontology was developed in Simon's group
<Susie> Bring RO uptodate and as close to usable for Larry as possible
<Susie> Tim: Use Tudor's model as the starting point, and assign folks to work on it
<Susie> Tim: Also take Larry's requirements into consideration
<Susie> Tim: Now we just need the happy volunteers
<Susie> Tim: Will volunteer Paolo as he isn't here
<Susie> Tudor: I will volunteer
<Susie> Larry: And some of Mike's time
<Susie> Larry: And maybe Elizabeth once she's finished her PhD
<Susie> Jack: My name should be on it
<Susie> Jack: Simon will argue that I shouldn't be on it!
<Susie> Tim: This is all subject to PI approval.
<Susie> Tim: Background information on esw wiki
<Susie> http://esw.w3.org/topic/HCLSIG/SWANSIOC
<Susie> Tim: There are 3 interest groups notes which provide good background
<Susie> David: Think myExperiment is somewhat separate
<Susie> Tim: Will put scientific discourse down as a stake in the group
<Susie> Tim: Can take up the discussion during the next call
<Susie> Tim: Any other open business
<Susie> ISWC workshop on scientific discourse