OpenOntologyRepository (OOR) Panel Discussion Session - Fri 22-August-2008    (1KWZ)

Conference Call Details    (1KX6)

Attendees    (1KL4)

Topic: Getting OOR Development Off the Ground: Identifying Synergies & Gaps    (1KY0)

Abstract    (1KY1)

by LeoObrst / MikeDean / PeterYim    (1KY2)

The OntologySummit2008 (Jan~Apr 2008) provided us with a great opportunity and platform to discuss pertinent issues relating to an Open Ontology Repository (OOR). We need now to move the OOR effort from discussion to action. This means that we need to identify usable programming code/services now and firm up implementation commitments - i.e., who can provide what now and in the future. This session will mark the kick-off of the OOR system development effort, during which we will bring together core developers and explore the synergies and gaps that we can make out of their collective initial contributions.    (1KY3)

We need to identify existing software that can be packaged into OOR modules and identify programmatic interfaces to existing repositories that can be used in a federation approach. If we could demonstrate the federation of existing repositories such as BioPortal, XMDR, SemanticMediaWiki, etc., that would be a major milestone. In addition, we want to ensure that Common Logic and its tools -- existing and emerging -- are included in the OOR realization (minimally, we envision OWL/RDF, CL, and possibly other languages/standards such as SKOS as being in the picture).    (1KY4)

Some key preparation guidelines primarily for the panelists, but also for participants and the [oor-forum] to consider, are the following:    (1KY5)

Refer also to details at the respective project homepages of the two initiatives at: OpenOntologyRepository & OntologySummit2008, and, in particular, the OntologySummit2008_Communique    (1KYB)

Agenda & Proceedings:    (1KYC)

Panelists' Presentation:    (1KYM)

Questions, Answers & Discourse:    (1KZE)

Questions and Discussion captured from the chat session:    (1KZN)

 PeterYim: Welcome to: OpenOntologyRepository (OOR) Panel Discussion Session - Fri 22-August-2008    (1KWZ)    (1KZO)
    * Subject: Getting OOR Development Off the Ground: Identifying Synergies & Gaps    (1KX0)    (1L19)
    * Session Chair:    (1KX1)
          o Dr. LeoObrst (MITRE; Ontolog)    (1KX2)    (1L1A)
    * Panelists / Briefings:    (1KLV)
          o Mr. MikeDean (BBN) - "A Modular Software Architecture for OOR"    (1KLW)
          o Professor MarkMusen (Stanford) - "BioPortal: An open ontology repository for community-based ontology management and review"    (1KLX)
          o Dr. LiDing (RPI) - "Search and navigate online ontologies"    (1KLY)
          o Mr. BruceBargmeyer (LNBL) - "XMDR Contributions"    (1KLZ)
          o Mr. BillAndersen (OntologyWorks) - "Common Logic: Tools and Prospects"    (1KM0)    (1L1B)
 RexBrooks: Question for Mike Dean: in terms of architecture for platforms, do we have recommendations for repository standards, e.g. ebXML Registry Repository? I ask because we could run into problems federating registries if their associated repositories don't support some features, such as repository custodian-specified classification schemes. So, might we need some base set of terms for the initial set of repositories and their associated ontologies?    (1L1C)
 Guoqian: Questions for LiDing: Does Swoogle support a triple store backend?    (1L1D)
 LiDing: Swoogle only maintain metadata, but does not use triple store to store content of ontology. However, Swoogle caches versions of ontologies, and selectively stores some triples in database. Moreover, semantic wiki does have a special triple store for semantic web ontology.    (1L1F)
 RexBrooks: Question for Bruce: Do you have a set of your own rules for taking UML to OWL?    (1L1G)
 LeoObrst: Question for Mark: Will BioPortal eventually provide more than 1-1  mappings? How do you envision that? Something like Smart/Prompt?    (1L1H)
 MikeBennett: Question for Bruce: In transforming UML XMI to OWL, are you following a standard transform like OMG's ODM?    (1L1I)
 LeoObrst: Question for Bill: Do you think it would be useful to provide semantic services to the OOR?    (1L1J)
 BillAndersen: Fabian suggests that IKL might be useful to consider as an upper bound for semantic translation.  Bill agrees.  As a practical point, Bill thinks that implementing Common Logic reasoning would be easier to implement.  CL has the additional advantage of being an ISO standard, although I don't think there's any problem with publiic accessiblity to IKL documents.    (1L1K)
 Guoqian: Is there bridge between SMW and Swoogle right now?    (1L1L)
 LiDing: not yet    (1L1M)
 Guoqian: are you planning to use Swoogle as a search engine against SMW?    (1L1N)
 HaroldSolbrig: Semantic Mediawiki has showed up in a couple of the talks.  Have the SMW (and Halo) been engaged directly in this effort?    (1L1O)
 HaroldSolbrig: Have the SMW (and Halo) *developers* ...    (1L1P)
 LiDing: RPI is currently investigating to build open ontology repository on semantic wiki (with halo extension).    (1L1Q)
 JohnGraybeal: Could SemanticMediaWiki (SMW) be considered as a front end to a common repository, or is it going to produce its own repository that may or may not interoperate with the federated OOR repository?    (1L1R)
 HaroldSolbrig: Response to John.  SMW has something that closely resembles a triple store on the back end, and is currently capable of producing triple-store output.  We are hoping to encourage and work with them to create a 'live' triple store base.  I don't believe that it would be that difficult.    (1L1S)
 LiDing: in general, SMW (semantic media wiki) seems to be a good web front end of maintaining ontologies (esp, collaborative editing ontologies). SMW can export its ontology in RDF/XML and then being indexed by OOR    (1L1T)
 HaroldSolbrig: SMW has also abstracted their back-end API (which SMW Halo is now using) which will allow multiple back end implementations.    (1L1U)
 LeoObrst: 2 questions for all: 1) Would it be useful to have analysis and transformation services for the ontologies in the OOR, e.g., to find out how to discover a Horn logic subset/transformation for a given ontology+rules?    (1L1V)
 BillAndersen: Instance of downward translation problem I discussed.  A good example is DLP.    (1L1W)
 LeoObrst: 2) Services to support rules, analysis of rules, mappings among rules?    (1L1X)
 MikeDean: plan to add a Translator module interface, indicating source, target, and completeness (logical and implementation)    (1L1Y)
 MikeDean: it's very useful to be able to chain these translators together    (1L1Z)
 MikeDean: Bill Andersen noted that it's very helpful if translators map back to original source as comments, etc.    (1L20)
 LiDing: In order to map back the original source, we will need some vocabulary to record the mapping track in metadata.    (1L21)
 BillAndersen: +1    (1L22)
 FabianNeuhaus: With respect of translating CLIF into formats which reasoners can use: it seems that the theorem prover community has some tools to convert CLIF into other formats http://www.cs.miami.edu/~tptp/    (1L23)
 LiDing: Fabian, RPI has done some work translating TPTP into PML, a semantic web proof markup language. http://www.cs.miami.edu/~tptp/Seminars/IWTPTP/Translation.html    (1L24)
 LiDing: TPTP and TPTS is a common language for annotating proofs generated by different inference engines.    (1L25)
 JohnGraybeal: We have 3 (at least technologies) whose interoperations are as yet somewhat unclear to me. Is there any plan to explicitly have this analysis/discussion?    (1L26)
 HaroldSolbrig: John - what 3 technologies are you referring to?    (1L27)
 JohnGraybeal: BioPortal, XMDR, SMW, and a few others already mentioned here.    (1L28)
 JohnGraybeal: Not to mention NeOn.    (1L29)
 HaroldSolbrig: Great question. It seems like there should be a common core w/ multiple views.  Q: Is there an architecture document for this implementation already available? If not, I think that architecture would be the obvious next step before we roll up our sleeves...    (1L2A)
 HaroldSolbrig: My own take would be an RDF triple-store at the core, but that may be biased against the CL approach?    (1L2B)
 JohnGraybeal: An overarching architecture into which these various components gracefully fit could be a very powerful way to start.    (1L2C)
 HaroldSolbrig: I'd be interested in participating and contributing to this.    (1L2D)
 BillAndersen: Re Harold's comment -- The use of RDF for meta-level would not bias against CL.  E.g. CL ontologies could be represented as RDF.  No problem there.    (1L2E)
 JohnGraybeal: Some principles for next steps could be: (1) Try to align with existing results.  (2) Identify overlaps and conflicts.  (3) Create an overarching architecture concept (e.g., "Backend Store" is a useful concept, perhaps) that could serve as a basis for identifying interfaces.    (1L2F)
 LiDing: next development meeting 29th or 5th ?    (1L2G)
 PeterYim: to leverage on the momentum, let's have our OOR-team meeting next week - Fri 2008.08.29 1.5 Hr. starting 9am PDT / 12 noon EDT / 16:00 UTC ... see you all there!    (1L2H)

Audio Recording of this Session    (1KZP)