Ontolog: Post OntologySummit2010_Symposium Meeting - Sharing and Integrating Ontologies - Tue 2010.03.16    (2C1P)

On-site Venue: "Employee Lounge" - NIST Building 101 (Main Building) ... (same venue as the OntologySummit2010_Symposium)    (2C4Z)

Conference Call Details    (2C1Y)

Attendees:    (2C2T)

Agenda & Proceedings:    (2C39)

Topic: "Sharing and Integrating Ontologies"    (2C3F)

Two or more application programs that interoperate successfully on common data must be based, explicitly or implicitly, on some agreement about the meaning of that data. Internally, those applications may use very different syntax, and some of their processing may depend on information that is not described in the common agreements. For example, a personnel database and a medical database may share information about the names, addresses, and Social Security numbers of many of the same people. But the business-related details in the personnel DB and the case histories in the medical DB would not be shared. In general, interoperability requires precise documentation of the syntax, semantics, and pragmatics of all the interactions among interoperable systems. Formal ontologies are metadata about the things, events, properties, people, and information involved in the design, implementation, and use of those systems.    (2C3H)

This report evolved from an email discussion in Ontolog Forum starting in February, 2010. Since many of the ideas were introduced, elaborated, and modified by multiple participants, itÂ’s impossible to credit any particular individual for any specific point. Instead, all participants in the thread with the subject line Foundation Ontology, Cyc, and Mapping should be acknowledged as contributors. Some related discussions, also starting in February, took place on the email list of the Architecture Ecosystem SIG of the Object Management Group, which also influenced the ideas presented in this report. Other publications and presentations are cited in the body of the report and collected in the bibliography at the end.    (2C3I)

Transcript of the online chat during the session:    (2C3L)

 see raw transcript here.    (2C3N)
 (for better clarity, the version below is a re-organized and lightly edited chat-transcript.)
 Participants are welcome to make light edits to their own contributions as they see fit.    (2C3O)
	PeterYim: .    (2CCP)
	Welcome to the Ontolog Post Summit Symposium Meeting 
	- Sharing and Integrating Ontologies - Tue 2010.03.16    (2CCQ)
	* Convener: Dr. JohnSowa (Vivomind Intelligence)    (2CCR)
	* Title: "Sharing and Integrating Ontologies"    (2CCS)
	Please refer to details on the session page
	at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2010_03_16    (2CCT)
	.    (2CCU)
	anonymous morphed into MattHettinger    (2CCV)
	anonymous morphed into DougFoxvog    (2CCW)
	MikeBennett: Could some ask that questione be repeated please, I can't hear them at all    (2CCX)
	SimonSpero: Gavagai?    (2CCY)
	SimonSpero: Undetached ontology parts    (2CCZ)
	PeterYim: my apologies about the mess with the chat-room links    (2CD0)
	MikeBennett: I was beginning to wonder if I was in the wrong room - thanks Peter!    (2CD1)
	MikeBennett: Re the last question (things which exist by virtue of being specified): one way in 
	which meaning is grounded in the business world is in legal systems.    (2CD2)
	SimonSpero: http://www.w3.org/2007/03/RdfRDB/    (2CD3)
	AmandaVizedom: The point on the current slide (titled "Consistency Check") is important. IME, most 
	folks developing ontologies to handle examples and/or instance-level data resist explicit 
	representation of instance-level examples, thereby losing this resource for testing...    (2CD4)
	AmandaVizedom: I'm referring specifically to projects in which the ontologies are used for semantic 
	metadata, for example, and the bulk of data remain in RDBs or other data sources. Here, there is 
	principled reason to keep the instances out of the ontology being developed ...    (2CD5)
	MikeBennett: @Amanda this is a very good point. One thing I hope to see developed later in this 
	conversation is the distinction between projects where the instance data is RDF/OWL individuals and 
	projects where the instance data is database instance data. So many issues require different 
	treatment in the two scenarios IMHO    (2CD6)
	AmandaVizedom: However, there is good reason to ontologize the examples (some sample data, some 
	sample messages, some sample service payloads). It's fine to do it in a distinct ontology, for 
	example, using import or inheritance mechanisms to make the ontology under development usable within 
	the example ontology. Now you've got something to do some meaningful testing on.    (2CD7)
	SimonSpero: We're getting deep in to Quineland here; I'd settle for non-monotonicity.    (2CD8)
	DougFoxvog: @Amanda Sample data can be placed in a model that uses a theory. Multiple models can use 
	the same theory, with the only difference being the instance data.    (2CD9)
	DougFoxvog: It is a good idea to keep instance data out of theory ontolgies, imho.    (2CDA)
	DougFoxvog: Some instances can be useful in theories. E.g., the Earth in a geographical theory, or a 
	legal code for a theory about how law applies to certain aspects of society.    (2CDB)
	MikeBennett: @Doug there will always be a need for certain instances in most class-level ontologies 
	(e.g. the USA, ISO etc.), but that's distinct from the sample data question, which I agree is an 
	important one    (2CDC)
	AmandaVizedom: @Mike Yes, I agree. In fact, this relates to our previous discussion of training 
	suitability to project. Much SemWeb-oriented training tends to assume that the instance data is 
	RDF/OWL, and to teach specific approaches (some elements of which we've discussed in last 2 days, 
	e.g. DL, no 2nd-order classes, no properties relating classes (as opposed to their instances), 
	etc....    (2CDD)
	MikeBennett: @Amanda - yes, to many sem-webbers the OWL/RDF web /is/ the uiverse of semantics. 
	People need to hear what John's saying about databases. All new, trendy movements assume a green 
	field site; the rest of us have to work with real world problems.    (2CDE)
	SimonSpero: The original web 0.9 took off because it integrated with all the data that was already 
	out there on the net    (2CDF)
	SimonSpero: It was a few years before http overtook gopher by traffic volume    (2CDG)
	SimonSpero: @Amanda, @Mike: the big problem with teaching OWL to people who know OOP is that 
	suddenly there's only monotonic inheritance    (2CDH)
	AmandaVizedom: @Mike However, many interoperability-driven projects, including mine and I think 
	yours, do not fit this. Rather, there are legacy data sources, not to be converted any time soon, if 
	ever, and the ontology is providing the explicit semantics absent from those sources (via markup or 
	indexing or wrapped services or...). For very good reasons, the sample data shouldn't be in these 
	ontologies. But we miss a much needed means of machine- or machine-assisted validation by not also, 
	separately, ontologizing some instance level data to serve as a test bed.    (2CDI)
	SimonSpero: @Mike, @amanda: the link I posted earlier is to a W3 workshop on mapping from RDF to 
	RDBMSes    (2CDJ)
	MikeBennett: @Amanda - indeed so. Some of the bright young things in financial services want to "do" 
	trendy SemWeb stuff, but most of them have real problems to solve. Since there's no merit in having 
	instance data in two places, it only makes sense for the ontology to be a business conceptual model 
	within a model driven stack of artefacts. But the test question is an interesting one, thanks for 
	flagging that up.    (2CDK)
	SimonSpero: @myself - and RDBMS -> RDF    (2CDL)
	MikeBennett: @Simon interesting link, it might help with some of the places where users of our 
	ontology are looking at ways to use it in solving real data problems.    (2CDM)
	SimonSpero: Mike: Best way to convince people that they don't want a jumbo triple store is to let 
	them build one    (2CDN)
	SimonSpero: Nothing like a giga-tuple table to slap some sense into the resistant    (2CDO)
	MikeBennett: @Simon re monotonic inheritance that explains why one sees ontologies with a single 
	hierarchy. I think there are interesting data mapping issues that require multiple inheritance in 
	the ontology mapping to distinct single inheritance data models across the organization or supply 
	chain.    (2CDP)
	SimonSpero: @Mike: multi is ok, but people want to override, because that's what they do when 
	programming    (2CDQ)
	AmandaVizedom: ...By doing so, we not only enable the kind of single-ontology checking John 
	described, but loads of potential additional testing, including testing of the implications of 
	particular alignments of ontologies, when such are needed for federated search, for example. Test 
	those alignments over test beds of ontologized instance-level examples that stand in for the 
	heterogeneous sources you aim to make interoperable.    (2CDR)
	anonymous morphed into AliHashemi    (2CDS)
	SimonSpero: Word & Object says we can't    (2CDT)
	AliHashemi: Sorry for being super late, was in a meeting till now...    (2CDU)
	MikeBennett: We're on Slide 10 (the 3D 4D question as an example of different theories)    (2CDV)
	AliHashemi: (thanks)    (2CDW)
	MikeBennett: This makes a lot of sense. I think in 4D anyway and was completely blindsided by the 
	fact that there are 3D theorists with their own peculiar definitions for continuants and the like.    (2CDX)
	SimonSpero: Perdurphiles    (2CDY)
	MikeBennett: It should be possible to frame a definition for "Continuant" which corresponds to what 
	John calls the Interface view - what it actually is, rather than how a 3D or 4D geek defines it    (2CDZ)
	SimonSpero: Are there individual rabbits, or are there just disconnected chunks of the unique Rabbit    (2CE0)
	AliHashemi: One comment about Slide 7 -- the lattice need not be a tree. There can be more than one 
	parent, and more than one root for any applied snippet of the "lattice of theories"    (2CE1)
	AliHashemi: I suppose the emphasized word is _like_ a tree    (2CE2)
	MikeBennett: <cheer!>    (2CE3)
	MikeBennett: That's re the questioner suggesting that these theorists come up with some real axioms 
	for their stuff    (2CE4)
	DougFoxvog: "Connected" in 3D and 4D have different definitions. The axioms do not conflict unless 
	they are using terms with inconsistant meanings.    (2CE5)
	DougFoxvog: Equating in 3D and 4D also have different meanings.    (2CE6)
	MikeBennett: Surely once we look at real axioms, one workaround that drops right out of real world 
	data is that there is a thing which exists over a period of time (howsoever modeled), and that thing 
	has a number of states and transitions between those states (again, howsoever modeled).    (2CE7)
	DougFoxvog: @Mike: what is considered to be a "thing" is a mental definition.    (2CE8)
	MikeBennett: @Doug surely the theorists aren't getting hung up on words just because some words may 
	have different meanings?    (2CE9)
	MikeBennett: @Doug good point    (2CEA)
	SimonSpero: Can we sum this up as saying hooray for empiricism?    (2CEB)
	SimonSpero: If there words could have two meanings, there would have been a sign on the dooor.    (2CEC)
	MikeBennett: @Simon: Philosophy Department (or is it?)    (2CED)
	SimonSpero: Enterprise architecture is basically enterprise archaeology    (2CEE)
	SimonSpero: Or forensic para-consistent epistemology : What the f*ck were they thinking?    (2CEF)
	SimonSpero: BTW, DICOM has 11 values for sex: 
	ftp://medical.nema.org/medical/dicom/final/cp373_ft.pdf    (2CEG)
	SimonSpero: Medical imaging quantised gender theory, because it had to    (2CEH)
	AmandaVizedom: IMHO, it would be of significant value were some research ontologists (i.e., those 
	for whom doing what follows would count as fulfilling the expectations of their positions, vs. those 
	for whom applied projects dominate) would pony-up with those axioms and proofs. Here's why: groups 
	of applied ontologists sometimes get into interminable debates, running years at times, over which 
	of two logically equivalent high-level representation approaches to use. In absence of agreement, 
	they also end up using each sometimes. Some will argue for equivalence and the practical importance 
	of picking one and moving on, but without the proof, this is rarely persuasive.    (2CEI)
	MikeBennett: @Amanda re our earlier (my email is down just now): Another twitter response has 
	@MikeHypercube methinks: "..Ontologies play a vital role in the broad Semantic Web Project vision, 
	the burgeoning Web of Linked Data .."    (2CEJ)
	SimonSpero: What's the #tag?    (2CEK)
	MikeBennett: @Simon in fact the world's first "non gender" person was declared in Aus in the last 
	couple of days. Cue database confusion.    (2CEL)
	SimonSpero: Mike: Gender != Sex    (2CEM)
	MikeBennett: :Simon The #tag for Linked Data is #linkeddata if that's what you're asking (apologies 
	if not)    (2CEN)
	SimonSpero: Oh - that tag    (2CEO)
	DougFoxvog: "We have reality" -- that's a theory. We have an "interface" to what we consider to be 
	reality.    (2CEP)
	DougFoxvog: Hopefully, there is some sort of agreement as to what "reality" is.    (2CEQ)
	AmandaVizedom: I could make a list ... but I won't. [Note re: Summit topic -- this would be nice to 
	cover in teaching as well: recognizing logically equivalent, or probably logically equivalent, 
	modeling approaches, and making choices -- either pragmatically or arbitrarily!    (2CER)
	MikeBennett: I think that there is a real case for a repository that identifies industry-led 
	standards (at a semantic level, where such exist), so that one can start to integrate the lattice of 
	actual, owned theories that are out threre.    (2CES)
	MikeBennett: With provenance metadata    (2CET)
	MikeBennett: Unified ontology?    (2CEU)
	AmandaVizedom: @Mike +1    (2CEV)
	SimonSpero: There's a workshop on ontology repositories at eswc in corfu this year    (2CEW)
	PeterYim: The "Shared and Integrated Ontologies (SIO)" project is born!    (2CEX)
	MikeBennett: Thanks Peter.    (2CEY)
	PeterYim: HUGE Thanks to *John* and All    (2CEZ)
	PeterYim: -- session ended 5:30pm --    (2CF0)
    -- end of chat session --    (2C3P)

Audio Recording of this Session    (2C3T)

For the record ...    (2C41)

How To Join (while the session is in progress)    (2C1K)