ppy/oor-metadata-07_chat-transcript_edited_20120424a.txt ---------------- Chat transcript from room: oor_20120424 2012-04-24 GMT-08:00 ---------------- [08:25] PeterYim: Welcome to the = OpenOntologyRepository: Metadata Workshop-VII - Tue 2012_04_24 = Topic: OOR Metadata Workshop-VI: Metadata for Financial Ontologies Session Chair: Professor MichaelGruninger (University of Toronto) Panelists: * Dr. OliverKutz / Dr. ChristophLange / Dr. TillMossakowski (University of Bremen, Germany) - "Ontology Links in the Distributed Ontology Language (DOL)" * Dr. JeromeEuzenat (INRIA & LIG, France) - "Ontology alignment metadata" * Dr. NatashaNoy (Stanford University, USA) - "Mapping repository in BioPortal" Session page: http://ontolog.cim3.net/cgi-bin/wiki.pl?OOR/ConferenceCall_2012_04_24 Mute control: *7 to un-mute ... *6 to mute (please make sure your own phone is not muted as well) Can't find Skype Dial pad? ... it's under the "Call" dropdown menu as "Show Dial pad" . == Proceedings: == . [08:27] PeterYim: Hi Michael ... [08:32] TillMossakowski: I just need to find my volume controls [08:32] anonymous3 morphed into JeromEuzenat [08:32] anonymous1 morphed into ChristopherSpottiswoode [08:33] ChristophLange: Hi, this is ChristophLange and OliverKutz [08:34] ChristophLange: yes [08:34] ChristophLange: in my room [08:35] ChristophLange: retrying, got an editing conflict [08:35] ChristophLange: slides now linked from the conference page, or here: http://kwarc.info/clange/exchange/2012-04-24-oor-metadata-mapping.pdf [08:39] anonymous morphed into Jeffrey Wallk [08:40] anonymous morphed into ToddSchneider [08:42] PeterYim: == Christoph presenting the work from OliverKutz, ChristophLange & TillMossakowski [08:43] PeterYim: the OntoIOp page (mentioned in Christoph's presentation) is at: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntoIOp [09:05] anonymous morphed into NatashaNoy [09:06] TerryLongstreth: Is the distinction between formal/informal equivalent to objective/subjective? [09:06] ChristophLange: Resolution for the DOL/OntoIOp terminology: call the general thing "alignment" (instead of "link"), with subconcepts "logical alignment", and "non-logical alignment" [09:06] TerryLongstreth: Or provable/unprovable? [09:08] TerryLongstreth: or demonstrable/inferred? [09:10] PeterYim: == JeromeEuzenat presenting ... [09:12] TillMossakowski: is there a language for comparing alingments, stating the joint conssistency of a set of alignments, and related things? [09:14] MichaelGruninger: what is the expressiveness of EDOAL? [09:15] TillMossakowski: it seems that EDOAL is for writing down alignments (including the use of complex class expressions), but not for comparing alingments? [09:17] TillMossakowski: what about using an alignments or a set of alignments for merging a set of ontologies? Is there a notation for this? In DOL, we have the combine notation for this, covering V-alignments, W-alingments, in general arbitrary colimits (and Hets can compute them). [09:18] TillMossakowski: in OntoIOp/DOL, alignments are first-class citizens that can be annotated with metadata [09:19] anonymous morphed into ChristopherSpottiswoode [09:21] anonymous morphed into seanmcbride [09:22] TillMossakowski: there is an interesting mapping of arbitrary (non-logical alignments) to pairs or quadruples of logical alignments (the latter then lead to V-alignments or W-alignments), see Jerome's V-alignment paper. [09:23] seanmcbride morphed into SeanMcBride [09:29] ChristopherSpottiswoode: Jerome: You say "Alignments are not ontologies" but they are surely instances of a very useful ontology? [09:30] TillMossakowski: unfortunately, I have to leave now, but I definitely listen to the recording of the audio [09:31] JeromEuzenat: Christian Meilicke's package: http://web.informatik.uni-mannheim.de/alcomo/ [09:31] PeterYim: == NatashaNoy presenting ... [09:33] JeromEuzenat: Till's #1: usually we interpret aligned ontologies as logics and run provers. In the case of Christian Meilicke, this was a simple translation. We had some other work (Chan Le Duc mostly) with our semantics: http://iddl.gforge.inria.fr [09:36] JeromEuzenat: Till #2: We have no notation for Alignment manipulation (only programs)... I have a PhD topic open on alignment algebra (and plenty of ideas) but nothing concrete. [09:39] JeromEuzenat: Michael #1: EDOAL complexity is horrible: it has full boolean + role-value maps + equations + inverse + cardinality constraints + individuals. I forgot, you may also add some pieces of programs. OK. The point is that we wanted to have a language that does not constrain us (and now we cannot translate many things into SPARQL...). [09:39] ChristopherSpottiswoode: Natasha, "Mappings in BioPortal areconcept-to-concept mappings" Maybe that's applicability is particularly strong in biological domains, where there is a vast base of agreed concepts? [09:40] MichaelGruninger: @Jerome: IS EDOAL equivalent to FOL? [09:41] ChristopherSpottiswoode: That vast base of agreed concepts I mentioned is what underlies Barry Smith's insistence on realism? [09:48] FrankOlken: Natasha, What are the most common upper ontologies in use in Bioportal? [09:51] NatashaNoy1: BFO, simply because it is an implicit requirement for OBO Foundary ontologies, so people include it [09:52] FrankOlken: @Natasha, Are you using a version management system to store the ontologies, ...? [09:54] MichaelGruninger: Question about specifying ontology mappings: Consider the set of mappings at http://code.google.com/p/colore/source/browse/trunk/ontologies/complex/combined_time/mappings/ The semantics of these mappings (which we call translation definitions in COLORE) is that the axioms of one ontology are entailed by the axioms in the other ontology together with the translation definitions. How are these sentences related to the ones specified in EDOAL? [09:54] JeromEuzenat: @Christopher (yes you understand that I am not a twitter), a that rate everything is an instance of an ontology... and it does matter. [09:57] ChristopherSpottiswoode: Thanks Jerome. That's an exciting field! [09:57] NatashaNoy1: @Frank: we have essentially implemented our own, very simple version management system [09:58] MichaelGruninger: retrying my conection [09:58] ChristopherSpottiswoode: (Jerome, I mean those meta-ontologies, not twitter!) [10:01] JeromEuzenat: @michael: lookink briefly, I would say that the only problem that come is that you have n-ary predicates. They can of course be simulated but this may not be the role of an alignment language. I have no idea of the relation with FOL. We did not even looked at the complexity because we did not wanted to write a prover but at some point this may become necessary. [10:04] MikeBennett: When looking at other formalisms, do you have a formal way to make sure that what is modeled in that formalism is real things as distinct from data elements in a logical model construct? Or does that not arise with the formats you are looking at? [10:04] JeromEuzenat: @michael: in addition, your stuff is a theory: you expect to cascade your definitions, while in alignments, the two sides of the "iff" are from two different ontologies. Again, expanding the definitions _may be_ doable in EDOAL (this depends on the complexity of your theory). Recursion may not be in EDOAL as generally as you may need it (maybe not in your examples). [10:11] ToddSchneider: If mappings are to be maintained in OOR and they cannot be represented in an ontology language, then additional mechanisms will be needed. Suggestions? [10:12] PeterYim: == open Q&A and discussion ... [10:15] ChristopherSpottiswoode: I strongly support Jerome's point about more expressive ontology languages simplifying mappings, in general of course. [10:16] MichaelGruninger: Interesting Issue: the relationship between the language used to specify the ontologies and the language used to specify the mappings between the ontologies [10:16] JeromEuzenat: My opinion: The _less_ expressive the ontology languages, the _more_ expressive the alignment language needs to be (in order to have given precision in the alignment). [10:16] MikeBennett: Is it practical to try and develop a common mapping ontology which covers all the requirements we have seen to date? [10:17] ChristopherSpottiswoode: Mike, That's certainly my position. Practical if done extensibly. [10:18] MikeBennett: @Christopher some of the work being done at the OMG (leveraging DC and SKOS terms as building blocks for this) may be a start to this. [10:18] MichaelGruninger: I don't think that we need a common mapping ontology; I'm not even sure that it makes sense to refer to a "mapping ontology", since we use the mappings to define the concepts of ontology by using the content of the other ontology [10:19] ToddSchneider: I would think it's more of 'categorical' construct: a product of languages [10:19] ChristopherSpottiswoode: Michael, such recursion should be no insuperable problem. Not if done correctly, that is. [10:20] MikeBennett: Why not? The relationships are themselves meaningful concepts? Our approach to implementing these is to them define them in terms of OWL annotation properties to annotate the actual ontology they are 'about' (we are doing this for provenance etc. and would do similar for cross reference perhaps). [10:25] ToddSchneider: Have to go. Deadlines. [10:26] ChristopherSpottiswoode: Many thanks to all speakers for their very valuable work and presentations! [10:26] PeterYim: it would be wonderful to continue exploring the incorporation of Jerome's Alignment Server into (the broader vision of) OOR as a value added service to ontology repositories ... Michael: consider another workshop to delve further into this topic [10:27] PeterYim: wonderful presentations ... great session ... thanks, everyone! [10:27] PeterYim: -- session ended: 10:27am PDT -- ----------------