ppy/oor-metadata-02_chat-transcript_unedited_20110513a.txt
PeterYim: Welcome to the
OpenOntologyRepository: OOR Metadata Workshop-I - Fri 2011_05_13
Topic: OOR Metadata Workshop-II
Session Chair: MichaelGruninger
see details on the session page
at: http://ontolog.cim3.net/cgi-bin/wiki.pl?OOR/ConferenceCall_2011_05_13
anonymous morphed into MariCarmenSuarezFigueroa
MariCarmenSuarezFigueroa: Bodil Nistrup
JouniTuominen: As a reference, in ONKI we describe ontologies with following metadata: http://www.seco.tkk.fi/u/jwtuomin/tmp/onki-ontology-metadata.ttl
JouniTuominen: When we designed the metadata schema, we weren't aware of the OMV. It would be reasonable to make it OMV-compatible, by using OMV properties when possible, and extending it when necessary.
MariCarmenSuarezFigueroa: do you have documentation about those metadata? Any document, or figure?
MikeBennett: Just a heads-up: we have started joint work with the OMG on Semantics Repository standardization; a major component is how to socialize the middle/upper ontology content. We would expect to leverage OMV terms.
MichaelGruninger: What metadata do we require to support modularity?
MichaelGruninger: What extensions to OMV are needed to support additional ontology representation languages such as Common Logic?
MikeBennett: We are still working out what is the best formal approach - at present we've got placeholder terms for most content, will look at how to align with external ontologies. In particular, not ontology-level terms but terms about terms (SKOS type of terms etc.).
TimDarr: This is the only relationship between ontologies that we would be interested in. We typically decompose our ontologies as taxonomies, relations, and rules.
TimDarr: A query, it seems to me, should match against the "composition" as opposed to the individual components.
PeterYim: [action] JouniTuominen will do a comparison of the ONKI metadata schema with OMV and report back (at the next workshop)
MichaelGruninger: @Tim: The modularity I was thinking about is along the lines of an ontology that combines a time ontology, a product ontology, and a business process ontology, which in turn extends a generic process ontology. How are all of these ontologies related to each other?
TimDarr: @Mike Is that different from the OMV import relations?
MikeBennett: THe kinds of ontologies we are looking at include UN-FAO, REA, XBRL (probably XBRL-GL), OWL Time, OMG's Time model,
TimDarr: OMV provides a way to list the ontologies that are imported
MikeBennett: Future one would include GeoNames, any ISO standards that are ontologies;
PeterYim: maybe we can take MikeBennett's use cases ... and see if OMV suffices (or what extensions are necessary to meet his needs)
MikeBennett: Units of Measure
MikeBennett: Action on me: Define Use Case
MikeBennett: @Michael at 14:33 this is very much our use case
MichaelGruninger: @Tim: In COLORE, we make distinctions between conservative and nonconservative extensions. In the latter case, some modules allow the entailment of new axioms
MikeBennett: I am working through an example with the REA ontology to see what approaches work best.
TimDarr: Use vs. extend, for example?
MikeBennett: I have a draft paper on that which might be useful.
TimDarr: Yes
TimDarr: That sort of ontology-to-ontology relationship seems to me to be useful: what you are doing with the ontology that you import.
MikeBennett: Our challenges are two-fold: The SR ontology adds decorative terms (Archetypes) which are not an OWL concept, and it disposes all high level terms under 3 top level lattices. So most (not all) external ontologies which have owl:Thing at the top, would be redisposed under the Sowa-derived lattice
MichaelGruninger: Would it make sense to focus on a particular domain or a particular set of ontologies such as time?
MikeBennett: Time ontology efforts: W3C time; OMG draft work; Mayo Clinic (Cui Tao, paper at ICBO in Buffalo in July); [http://www.fpml.org/ FpML] derived time terms; our own draft terms
MikeBennett: PatHayes: Catalog of Temporal Theories
MikeBennett: One thing we are thinking of is separating business domain specific (Legal, Accounting) versus domain neutral things like time and math and units of measure which are not related to any industry
MariCarmenSuarezFigueroa: Related to TimDarr's comment: with OMV you can describe the relationships between two ontologies (imports, versions, compatibility); and if more are needed OMV can be extended.
PeterYim: In the interest (and urgency) of our need to stand up a "production" instance of the OOR ... noting that we only have a "sandbox" instance of OOR now, and that the key differentiation between the two is the implementation of "Gate Keeping" which, among other things, *requires* anyone who uploads an ontology to OOR to populate the metadata (which is not required for the OOR-sandbox - Question: can we just go ahead and implement the "Gate Keeper" with OMV (as is) or (we here, take the time to) specify a "minimal required subset" ... can we discuss this?
MikeBennett: Tree analogy: mid level is under-specified. things like legal, country, legal entity, are owned by some industry but are terms which other industry need to refer to (financial: securities are contracts; insurance: risk etc.
MichaelGruninger: @MikeBennett: Which of the ontologies that you mention in your chat posts can be submitted to OOR Sandbox?
MikeBennett: @MichaelGruninger the thing that can be submitted at present: OMG time work when this is done. The Mayo work definitely. Need to deduplicate these two
MichaelGruninger: Action Item 1: Focus on time ontologies as a testbed for exploring relationships between ontologies
KenBaclawski: The OMV would have to be extended to deal with the gatekeeping metadata such as the owner, group, status, rights, etc. I noticed that the onki ontology has some properties dealing with these issues.
MariCarmenSuarezFigueroa: Metadata such as status, developers, rights, etc. are also covered by OMV. We can analyse which ones from ONKI are not covered yet by OMV.
MikeBennett: Accounting: XBRL and REA which have related but different terms and ontological commitments
TimDarr: How about something like SPIN?
MichaelGruninger: Action Item: Contact GaryBergCross (SOCoP) to contribute geospatial ontologies to OOR Sandbox
MikeBennett: Geographical ontologies would be a very useful area to align and cross reference ontologies (for instance UN-FAO has very specific scope; ISO 3166 has no axioms; GeoNames?...
MikeBennett: Geographical is a good one to focus on because it is more like an "industry" vertical (there are expert bodies, standard bodies etc. specific to it) but it is a set of terms which almost any ontology will need to refer to (securities; buildings management; etc.)
MikeBennett: And if anyone wants to be involved in the OMG joint work drop me a mail
PeterYim: Next session: Fri 5/20 "Architecture & API - V" ... then Fri 5/27 "Metadata III"
PeterYim: -- session ended: 7:01am PDT --