Personal tools
You are here: Home Meetings & Workshops 1st meeting Meeting Notes

Meeting Notes

by schild last modified Oct 07, 2009 06:30 PM

Meeting notes from the first SONet community workshop, Oct. 7-9, 2009

Oct. 7 summaries

* = action items

After presentations from the O&M, Seronto, ODM, OBOE, VSTO, and DarwinCore+ communities, we see a lot of congruence among models for storing observational data.

* An initial task for the SONet postdoc would be formally mapping among these different models, to determine compatibility.  Develop white paper and share.

Important task is to scope the core model.  Are we talking about "earth science" data; are spatial issues always paramount? what about lab data.  Elaborate on the "where" part. Similarly, how important is modeling the temporal dimension, and mixed cases. On ODM, timescale and time support is critical-- continuous, instantaneous, gaps, periodicity. Biological taxonomies?

Among different observation models, might need best practices to clarify use and resolve some of above questions.

Ownership/governance-- who should be developing models? what are definitive sources for information.  E.g. * Units-- NIST, SI--contact them to find out status?  E.g. SWEET-- non-monotonic revisions can be problematic to application builders; outreach.  E.g. OGC O&M-- will be ISO standard-- that has 5 year revision cycle.  Stability of model can be helpful, but also limiting if isn't timely opportunity for community to comment/suggest.  This isn't as big a concern if the core model can be extended; but deeper problems could exist in core model that need immediate addressing.

OWL got rec status within 3 yrs from W3C.

Latest draft of O&M has XML schema and UML implementations.  Also possible to convert to OWL.  OGC maintains a schema repository.

Need to continue developing scenarios for whether a given model covers some specific type of data-- e.g. time-series, experimental treatments, cell-level vs table-level resolution, output from models --> use cases

Need formal representations of models.

Who is the target audience?  Is this a processing system or data exchange system?  Who do we want to adopt this?  Is effort for humans or machines?  Is for both-- must be machine interoperability layer, but ultimately audience is scientists.  How much do we expect machines to automatically locate and integrate resources?

Should we be developing different products for different audiences:  one Core model with lots of  domain ontologies (domain scientists involved), or specialized observation ontologies for different domains?  For Core model, is mainly tool providers, but must be accessible to domain scientists to contribute. But domain ontos require more direct involvement.  Then, leads to governance issues about managing and versioning.

In order to integrate across data sets, must involve some markup/annotation to expose common connection points.

Want scientists to be able to semantically enhance their data. Deborah often has done this with a computer scientist working with domain scientist.  Ultimately, in genomics (GO) and biomed/cancer groups, scientist can autonomously annotate and leverage ontos, but only after years of involvement with KR experts.  Over the years, GO went from complex to apparently much simplified-- deep subsumption, but few properties.  At one point had lots of expressiveness, which now gone.

Spatial scaling issues.

We can't deal with data mis-use issues.  We just need sound ontologies that leave it up to end-user.

For Use cases, choose one that some group has already manually architected the solution.  Now we try to re-solve and see if make more efficient.


 USE CASES: leadership team discussion

Want Use Cases to inform core model requirements

Use Cases to provide basis for Interoperability challenge

*Identify services that can be used to support Use Cases (for demo projects)

Wikipedia Templates for Use cases? 

Setup plone accounts for scribes--  Matt, Huiping, Shawn

  * types of queries

  * type of integration needs scientists want

  * cross-domain integration

  * range of data types

  * identify services out there with suitable data, also collect data

1) Oceanography/Marine Coastal
Luis *, Matt, Margaret, Jessie, David, Ben, Deborah


2) Urban socioeconomic/climate/landscape

Corinna*, Huiping, Rob, Simon, Nick, Peter


3) Citizen sci/biodiversity/range/traits

Mark*, Shawn, Steve, Damian, Benno, Flip,













Document Actions