ppy/symposium-day-1_chat-transcript_unedited_20120412a.txt ------------- Chat transcript from room: ontologysummit2012 2012-04-12 GMT-08:00 ------------- PeterYim: Welcome to the = OntologySummit2012 Symposium - Day 1 - Thu 2012-04-12 = Summit Theme: OntologySummit2012: "Ontology for Big Systems" Ontology Summit 2012: General Co-chairs: Dr. LeoObrst & Dr. NicolaGuarino Ontology Summit 2012: Symposium Co-chairs: Dr. RamSriram & Professor MichaelGruninger Session page: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012_Symposium Mute control: *7 to un-mute ... *6 to mute Can't find Skype Dial pad? ... it's under the "Call" dropdown menu as "Show Dial pad" . == Proceedings: == . [07:22] FrankOlken1: Mary Brady is speaking. [07:24] AliHashemi: communique viewable here: http://ontolog.cim3.net/OntologySummit/2012/communique.html [07:24] ToddSchneider: Nicola, I think I understand your comment about 'human in the loop' assertion that was in the communique. I think the issue is what is identified as the 'system'. Your interpretation seems to include the larger environment in the which the/a 'system' operates. Is this correct? [07:25] NicolaGuarino: For the record, I state what I just said: In a previous version of the communique, there was a statement saying that "while for smaller ssystems it may be viable to have a human "in the loop", the scale complexity and costs of big systems or big data often excludes such an option". I believe that this is not the case. The more complex systems are, the more people must be (and actually are) taken in the loop. Consider for instance the pilot and the passengers of the plane mentioned by Henson. We need to develop models which DO take these humans in the loop. The crucial thing is that, despite the fact that the pilot can be described in terms of his/her expected behavior, still can decide to behave differently... My bottom line is that modeling complex systems requires being aware that such systems are intrisically socio-technical-economic systems... [07:26] AliHashemi: @Nicola - Agreed, the phrasing was poor. I think the point that was garbled is that it is less viable to have manual modfication and updates in a complex system, not that humans ought not be in the loop. [07:26] SimonSpero: [Question for Mary that I will bring up later: open question as to whether it is subject domain experts with some data management training, or computational scientists and information scientists with some domain training] [07:27] MatthewWest: @Todd: I agree with Nicola, and made a similar comment in my review of the Word version. [07:29] ToddSchneider: Nicola, Matthew, We've modified the offending sentence from the communique. [07:30] MatthewWest: @Simon: In ISO 15926 we use both of these. Most curation is done by domain experts with ontology training, but that is overseen by ontology experts who pick up any issues. [07:32] SimonSpero: Matthew: A friend of mine just defended her dissertation comparing IS and domain specialists approaches to data management and metadata generation; the results are interesting [07:32] GaryBergCross: One implication of the Nicola's point on intrisically socio-technical-economic of systems & complex systems is the need to have ontologies for the social-intentional-goal aspect of designed systems. DOLCE has useful work in this direction. [07:32] SimonSpero: Curation happens long after the funding has stopped. [07:37] SimonSpero: [Evaulation Paul Cohen, call your agent- http://www.amazon.com/Empirical-Methods-Artificial-Intelligence-Bradford/dp/0262032252 ] [07:37] MatthewWest: @Simon: Curation has to be part of "business as usual". You need to establish a quality management system around your ontology (which will often be called Master Data). [07:39] SimonSpero: Matthew: Curation as applied to big data has to be part of the fabric, but is much bigger scale than just ontology [07:44] NicolaGuarino: @Ali: I've seen you are working again on the Google doc version of the communique. Is it the latest one? [07:46] anonymous1 morphed into RaviSharma [07:46] AliHashemi: @Nicola, yes. [07:47] AliHashemi: (it has not yet fully incorporated your revisions from this morning) [07:48] NicolaGuarino: @Gary: Indeed. This is all related to so-called social ontology , mainly inspired to Searle's work. On the ontology (and in general, the philosophy) of technical and socio-technical artefacts, I would recommend Vermaas, P., Kroes, P. & Franssen, M., 2011. A Philosophy of Technology: From Technical Artefacts to Sociotechnical Systems, Morgan & Claypool Publishers. [07:54] SimonSpero: HighFleet ECLIF has time, based on IKL, think [07:55] Kyoungsook Kim: Based on Big data, the ontology also becomes large-scale, dynamic, and diversity I think. How about managing ontology itself? [07:56] NicolaGuarino: @Todd: I am not aware of current work aiming at bridging the gap you mentioned (may be there is some, though). Of course, mathematical models are definitely MUCH better for certain things! The interesting thing would be being able to point, say, to the variables appearing in a mathematical formula, explaining what they are supposed to represent in terms of an ontology... Is anybody aware of anything like that? [07:57] GaryBergCross: @Nicola Social ontology's role might be a good topic for a future Summit, or at least a track of one. [08:02] FrankOlken1: The summit is scheduled to resume at 11:15 AM Eastern Time. [08:04] RaviSharma: Mary - the HDF EOS was designed by Mike and some of us for Earth science datasets that included metadata attributes - how semantics and ontologies such as thru RDF and its value addition assesment would be great to know. Further, connected sysytems that use this big-data from active archives would greatly benefit from semantics enrichment that was not in original metadata as they could not conceive all further uses of datasets from NASA and NOAA. [08:15] PeterYim: == reconvening (after the coffee break) [08:16] PeterYim: == Track-4 summary report ... SteveRay presenting [08:16] NicolaGuarino: @Gary: I was thinking exactly the same :-) [08:17] MatthewWest: @Simon: Generally curation of the "big data" should be minimal as long as it is captured correctly (against an ontology) in the first place. What problems are you finding? [08:20] SimonSpero: Matthew: Formats may change; hardware becomes obsolete; media may fail; cryptographic primitives may be cracked; transformations may be lost; provenance may be lost or over generalized, etc [08:21] DeborahMacPherson: Can someone please provide a list of recommended reasoners - related to Steve's bullet "ontologies are not always applied to enable reasoners - sometimes just as a more rigorous data modeling approach." [08:24] GaryBergCross: @Deb There are, of course, some reasoners in a tool like Protege. [08:25] SimonSpero: RaviSharma: HDF5 has facilities for attributes - http://www.hdfgroup.org/HDF5/doc/UG/UG_frame13Attributes.html [08:26] MatthewWest: @Simon: I recognise those. For process plant you need to curate data over 30-50 years - the life of the plant. These are standard problems for data management in general. No easy answers, you just have to have a continuity plan, or face the consequences. [08:27] DougFoxvog: @Deborah: The elephant (in the room?) for reasoners would be Cyc. Protege has a little bit of reasoning. I don't have a list of in between reasoners at my fingertips. [08:27] GaryBergCross: @Nicola Put me down for help organizing the Social ontology's topic as a topic for a future Summit. [08:27] SimonSpero: Matthew: The company archivists will need the data possibly for longer than the life of the plant [08:27] SimonSpero: But agree with your point [08:28] ToddSchneider: So what is an ontology pattern? [08:30] SimonSpero: Deborah: Reasoners for OWLish things (open source division): Pellet (C&P) - http://clarkparsia.com/pellet/ :: Hermit (Oxon) - http://hermit-reasoner.com/ , Fact++ (man U) - http://owl.man.ac.uk/factplusplus/ [08:31] GaryBergCross: I like having lists of needs at any given moment and some benchmark status. It would be nice to revisit these in few years and see what, if any, progress has been made. [08:31] CoryCasanave: @Tod - patterns may be another treatment of the need for composites [08:31] DougFoxvog: @Todd: It seems an ontology pattern is a small ontology with a set of relations. When you develop your own ontology, you import the pattern, giving you the relations and a structure under which to put your concepts. [08:32] DeborahMacPherson: Thanks everyone - helpful [08:34] MikeBennett: Ontology patterns are a lot like what we have defined as archetypes I think - only more general if anything (the whole/part example). In general it makes sense that for anything that is a kind of thing, you can capture what are the minimal and neccessary facts that make it such a thing. When we did this, some (Address comes to mind) are more general than the terms you might find in specific industry ontologies. [08:34] TerryLongstreth: Hans Polzer at the mic [08:36] RaviSharma: Steve - do you include ontologies for genetic drug development in clinical-genomics or just hospital practice and Medline or SnowMed etc? [08:38] CoryCasanave: @Todd, in terms of dumbing it down for users, it is frequently the other way around. You have to dumb down what users want to say about their environment and systems such that it is expressible within the limits of many of the formal languages. If concepts expressed in an ontology dont have a counterpart for some stakeholder, what are they expressing? [08:38] RaviSharma: Steve - is tghere an understood path from legacy to semantic transition that also includes SOA as the main usage for modernizing Legacy, i.e. should there be semantic SOA as was proposed by USAF sometime ago? [08:41] MikeBennett: @Cory +1 - it's usually not users we need to dumb things down for, it's the data geeks. They just think they are smarter because they know a particular thing the users don't, and also often lack awareness of what's in others' minds. [08:42] FrankOlken1: Is this Amanda Vizedom speaking? [08:42] GaryBergCross: Yes [08:42] LeoObrst: Yes, Frank. [08:44] GaryBergCross: One should consider efforts to develop controlled vocaabularies that can be faithfully converted to languages like CL or OWL. One example is Rabbit developed for expressing OWL. [08:48] RaviSharma: Amanda - can you use interdosier buy-in? eaocabularies in their domain of expertise, that way you getmain vocabulary translations or transformations to relate to the experts, namely use techniques that you have developed similar to enterprise vocabularies only in this case present them in their familiar v [08:48] RaviSharma: Amanda- continuing ..vocabularies and get quick buy-in? [08:48] SimonSpero: GaryBergCross: Rabbit is not the ideal CNL; Attempto may be a better solution - http://attempto.ifi.uzh.ch/site/tools/ [08:49] PeterYim: == AmandaVizedom and MikeBennett presenting the Cross-Track-A1: "Ontology Quality" Summary [08:50] GaryBergCross: Aldo Gangemi describes Content Ontology Design Patterns as small ontologies that mediate between use cases (problem types) and design solutions. [08:50] SimonSpero: I meant what I said and I said what I meant - a model is faithful 100% - Horton hears an OWL [08:52] SimonSpero: Mike - ISO 9001 ; how do you create a process so that all ontologies produced are wrong in the same way [08:52] SteveRay1: @Ravi: With respect to your question about different kinds of biomedical ontologies, I was not trying to describe a categorization of such ontologies, but only reporting on the specific ontologies that had been presented. [08:53] LeoObrst: @Cory and Mike: yes, if you over-simplify the semantics of the users to fit your formalism's expressiveness (or comparable reasons), this is a kind of semantic shoe-horning. This also comes up when you have multiple communities and you define a common, intersecting semantics. If there are constructs in that intersecting semantics that obliterate real distinctions particular communities make (and make differently), users will balk. Surprise. They may agree formally, then just not use that so-called common semantics. This comes up all the time when developing Community of Interest vocabularies and ontologies. [08:54] MatthewWest: Are we still on slide 2? [08:54] AliHashemi: slide 6 [08:54] MatthewWest: VNC is showing 2 [08:55] SimonSpero: pokes VNC [08:55] GaryBergCross: These problems remind me a bit of the early AI and Expert Systems days where you expect the "expert" to explicitly know what they are expert on and are capable of articulating that knowledge. [08:56] SteveRay1: @Ravi: Regarding transforming legacy systems to ontology-based systems, I was speaking primarily about transformation of information models into more expressive semantic models. The question of system architectures is more complicated than that (in my opinion), and I don't have an informed opinion about transformation from "legacy architectures" into "semantic architectures". [08:57] SteveRay1: @MikeBennet: I agree which is why I come down on the side of teaching domain experts a little semantics rather than trying to teach ontologists a little domain expertise. However ultimately it needs to be a team activity. [08:58] SteveRay1 morphed into SteveRay [08:58] SimonSpero: GaryBergCross: There is a pretty universal rejection by practitioners of KK - that's why the issue of designing tests (especially automatable tests) is hard; experts are only experts with sufficient context [08:59] GaryBergCross: Was what is meant by "semantic processing functions" well described in the session? [08:59] SimonSpero: Otherwise the answer is "it depends" [08:59] TerryLongstreth1: How do you define comparative criteria for evaluating degree that an ontology or any part approaches desired quality? [09:00] GaryBergCross: @Spiro by KK did you mean KR? [09:01] SimonSpero: GaryBergCross: KK - Knows and Knows that knows [09:01] DougFoxvog: @Leo: the problem of developing a standard vocabulary with different communities involves conflicts because the different communities use the same term differently. This need not affect ontologies. One can model the concepts of each field as each field views them & inter-relate the concepts. You can use different namespaces, which would make clear that FINANCIAL#Bank is different from BILLIARDS#Bank is different from MIL-STRUCTS#Bank. [09:01] GaryBergCross: Ont quality might be a point measure or something that is like robustness that holds up over time and over uses. [09:02] GaryBergCross: @Spiro thanks, now I know. Sort of a metacognition idea (MC) [09:02] SimonSpero: http://en.wikipedia.org/wiki/Epistemic_logic [09:05] LeoObrst: @Doug: yes, that is in fact what you need to do, i.e., map different vocabularies to common ontologies, so that different terms actually map to same or similar concepts (with the filagree of meaning preserved). [09:06] GaryBergCross: I'd like to think that there is increasing agreement on quality in the community, but perhaps the understanding of it is still growing and changing. [09:08] RaviSharma: Amas coverage - similar to code coverage while testing software?nda - I am not aware of any formal theory or use case measure of completeness of requirements. Can ontology including semantics and vocabularies dermine extent of requirement [09:10] RaviSharma: Amanda - continued....Amanada re-writing the comment - is there a formal way through use of ontology or vocabulary and semantics of determining coverage of requirements similar to code covergae thru test tools? [09:10] CoryCasanave: @Ravi, would 100% principle of [ISOTR9007] help? [09:10] DougFoxvog: Also, we need to be able to separate the natural language terms used from the name of the entities created in the ontologies. We should be able to accept multiple words and phrases as signifying the same entity in an ontology. [09:11] RaviSharma: Cory- does that foormalise when requirements are met, earlier were only good use cases generation that are tested againt implemented sysytems [09:12] CoryCasanave: @Ravi, it essentialy states that 100% of the requirements of a system are in a "conceptual model" [09:13] RaviSharma: Cory - but whatdetermines that stated requirements are initially or even finally cover all requirements (completeness of requirements)? [09:15] RaviSharma: Todd can you read my anonymous interpretation of my long comment on the communique? [09:15] CoryCasanave: @Ravi, not sure - but I think it comes from making sure that everything in an "implementation" are derived from the conceptual model and that any new requirements are stated thru revisions of that model [09:16] LeoObrst: @Doug: Yes, the labels actually used to name the concepts in the ontology really are arbitrary symbols (but one should pragmatically make those as human-meaningful as possible). The terms in the vocabularies that map to those "nodes" can be as as simple or complex as one needs. [09:17] CoryCasanave: @Leo - if they are arbitrary why is there so much discussion of them? [09:21] GaryBergCross: On this idea of "proper" ontological classes allowing additivity there are theories of semantic "blending" that suggest it is a bit more complex. See Fauconnier, Gilles and Turner, Mark . 2002. The Way We Think: Conceptual Blending and the Mind's Hidden Complexities. Basic Books. [09:22] LeoObrst: @Cory: yes, the problem is that people will fight for their "label", i.e., their words, because they conflate their vocabulary terms to what those terms actually mean (i.e., the concepts). If you can preserve both, they will be happy. [09:23] RaviSharma: todd - please reply to my communique comment ananymous? [09:25] DeborahMacPherson: @Leo and Doug - any recommendations or examples regarding best path forward when a common ontology is "wrong" compared to one or more authoritative vocabularies? [09:25] FrankOlken1: Revelytix is a major contractor to DOD for their semantic web work. [09:26] RaviSharma: Amanda - use cases- completeness that requirements are met - for many use cases states that implemented sysytems are aligned to requirements. [09:28] RaviSharma: Amanda -Ontogies are to be used to validate models and also requirements, can they also depend on use case validation or will there by a loop recursively between ontology and sysytem implementation models? [09:28] SimonSpero: If an ontology entails certain consequences that native speakers of the vocabularies reject, then the mappings of those vocabularies to the ontology cannot be correct. So the mapping must be changed so that the incorrect conclusions are not drawn, or the ontology must be altered. [09:28] GaryBergCross: @Deb Just to add to this discussion o "wrong" ontologies. I can imagine that we would look at 2 ontologies that may differ in several ways with one seeming to have quality in one area while the other has more "quality" elsewhere. We might be able to say that one is "wrong" in one area, but often it might be an issue of useful for what purposes. [09:31] MatthewWest: What slide are we on? VNC is on 4 [09:31] AliHashemi: slide 7 [09:31] SimonSpero: GaryBergCross: There are many ways for an ontology to be right. However, there are infinitely more ways of being wrong. (Quine 1960 , 2) [09:31] RaviSharma: Cory - like John Sowa says I think - relational databases and semantics / ontology have to interoperate with eacch other but how and how well and also easily understood by practitioners? [09:32] DougFoxvog: @Cory, ref: Leo. Another issue is that the interface to ontologies is so often code. People look at the names of the entities in the ontologies. Users don't normally do this with code. So, there is no big argument with what the correct name of a Java method should be. Coders just use the name, whatever it is -- they don't care. If one normally used an interface to an ontology, which showed a term (or set thereof) tailored to the user's context, it would be less of a problem. [09:33] RaviSharma: Todd- Cory rearticulted my communique comment more succintly in his talk but kindly capture that in the communique. [09:35] DougFoxvog: @Deborah: At NIST, we have run into errors in commonly used ontologies, e.g. a standard ontology of cell types had several instances of groups of classes that were subclasses of both "plant cell" and "animal cell". These resulted from someone deciding that a general term (epithelial [09:35] GaryBergCross: In discussing "federation" I start to think of different part-whole relations and of components of a system. There are probably several useful distinctions, some maybe new, that might be made for discussing types of relations in a Federation. [09:36] DougFoxvog: cell or stem cell) were types of Animal cell (since that was their context) even though they had already been defined with both plant & animal cell subtypes). [09:36] SimonSpero: @DougFoxvog - Drew McDermott noted problem with code too - AI meets Natural Stupidity : http://www.csee.umbc.edu/courses/471/papers/mcdermott76.pdf [09:37] DeborahMacPherson: @Doug - how did you straighten this out? [09:38] DougFoxvog: @Deborah (cont.): So what's the best path forward when a common ontology is "wrong"? In this instance we informed the party responsible for the faulty ontology. In general, we haven't yet created a method of dealing with it. [09:39] DeborahMacPherson: ok - so contacting the person is best. Seemed best [09:39] SimonSpero: [Part -Whole: tool to build mereologies in OWL : http://www.meteck.org/files/ontopartssup/supindex.html [09:39] GaryBergCross: @Nicola "authoritative reference ontologies" is another good topic for a future Summit. [09:40] MatthewWest: @Doug: Yes, that is a general problem with ontologies developed and used by communities. You need a good bug fixing/quality management approach to continuously improve the ontology. [09:42] DeborahMacPherson: continuous, versioned, cyclical, traceable improvement [09:43] DougFoxvog: We are looking into methods for analyzing ontology quality, both at the logical level (such as the presented case) and at the level of coverage and complexity (e.g., is it just a taxonomy; what types of relations does it have; & many other features). We are looking at published methods (including OntoClean [NB: this is not an official recommendation of that method]) and modifications thereof. [09:44] GaryBergCross: John Sowa's the Glossary discussion of "Building, Sharing, and Merging" (of ontologies, but could perhaps be systems) probably makes some useful distinctions including aligning vs integrating. http://www.jfsowa.com/ontology/ontoshar.htm [09:46] DougFoxvog: @Simon: there are ontology of software bug types out there. I haven't read the McDermott paper (at least in the last 30 years). [09:46] NicolaGuarino: I have to go right now... [09:46] MatthewWest: @Doug: Yes Ontoclean is an approach. Unfortunately it makes ontological commitments which means it has less value for evaluating ontologies that do not share those commitments. [09:47] MatthewWest: I can't make the session after lunch. Apologies. [09:47] DougFoxvog: @Matthew. I totally agree. [09:47] MikeBennett: Is there a way of articulating ontological commitments themselves? As a feature of the requirements for an ontology, being able to articuate these formally would provide one more dimension of ontology quality. [09:47] AliHashemi: Re plug'n'play - I think there's quite a broad spectrum. In the sense of say, MSWord being plug'n'play to being able to use software libraries. [09:48] RaviSharma: Peter - adherence to standards is the key to plug and play and APIs do not allow that. Software such as XMI does not work with each other so ontology standards ahve hope for future plug and play. [09:48] AliHashemi: Many vendor solutions that try to be general still require substantial efforts to customize them to various customers. [09:48] PeterYim: == we are breaking for lunch now [09:48] AliHashemi: plug'n'play at the level of "word" seems quite a far ways off.. [09:49] DougFoxvog: @Mike: there should be a way of articulating ontological commitments. I haven't seen much work on this. Cyc has a method for stating essential facts about its microtheories -- but it is not strong enough for what you want. [09:49] MatthewWest: @Mike: I certainly think that being able to state the ontological commitments of an ontology is a key measure of ontology quality, and is the key to understanding the capabilities and limitations of an ontology. [09:58] vnc2: == we'll be resuming at about 1:45pm EDT [10:00] vnc2: please note that remote participants may have to call in again for the afternoon session (not sure, though ... it's the same phone number and PIN, but nonetheless, a different conference bridge session.) [10:52] PeterYim: == afternoon session starting - 1:52pm EDT [11:04] MikeBennett: @Doug @Matthew thanks, that sounds like a fruitful area for further research. Maybe building on what Cyc has. [11:14] FrankOlken1: Silica based life form??? [11:14] FrankOlken1: Can speakers identify themselves please? [11:16] SteveRay1: That was Simon Spero [11:16] SimonSpero: Not me [11:18] GaryBergCross: @Frank Sponge-like creatures used biomineralization that included Silica for filtering, before we see hard shells of CA etc. [11:19] SteveRay1: @Simon: Sorry, you are right. It was Jim Disbrow [11:19] GaryBergCross: Do we ever use references to support one or more arguments? [11:36] PeterYim: @Todd & Ali - the EliotSiegel "Dr. Watson" scenario was presented at: EliotSiegel [11:37] PeterYim: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2011_03_24 [11:48] AmandaVizedom: Regarding Ontology Quality section & what MikeBennett just said, see our synthesis page http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012_Quality_Synthesis [11:57] Rex Brooks: If we want to be more politically correct with "should" we could adopt RFC 2119. [12:02] SimonSpero: @RexBrooks: that's "SHOULD", not "should" [12:03] Rex Brooks: Yes, if we adopt 2119 it would be SHOULD. I left is lowercase in quotes to highlight that. For now, we are using dictionary definition only. [12:05] Rex Brooks: But even if we adopt RFC 2119, SHOULD is narrowly defined and is not a strict conformance requirement. So even then it is a bit of waffling. However, I would prefer adopting conformance language as in RFC 2119. [12:09] SimonSpero: @RexBrooks: right -the deontics are hard to specify formally, since SHOULD and MAY are both \diamond. You are supposed to have a plausible excuse for ignoring a SHOULD, but it's not supposed to cause interop fails. [12:10] DougFoxvog: +1 Ram, the first line in the Recommendations section should clarify that the Ontolog Summit is making these recommendations. Also, as i commented, it should be pointed out that a summary is being given before the details. [12:13] SimonSpero: BOGROT BLUF [12:14] TerryLongstreth: On the recommendations- I'd suggest the authors adopt a simple rule for fostering consistency across the summary bullets; perhaps each statement have a clear espression of the intentended recipient of the rec. [12:15] DougFoxvog: @Simon ???? What is "BOGROT BLUF"? Were you touch typing with your fingers on the wrong keys? [12:16] RaviSharma: Todd - please llok at the link and my comments these capture results from 40 references.http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012_BigSystemsEngineering_CommunityInput [12:17] SimonSpero: BOGSAT BLUF would be better (bunch of guys sitting at table) (bottom line up front) [12:18] PeterYim: == we are taking a half-an-hour break now ... resuming at 3:45pm EDT [12:18] FrankOlken1: Summit will resume at 3:45 PM EDT. [12:47] PeterYim: == session to review the communique is resuming now ... [12:51] SimonSpero: http://www.cadalyst.com/management/what-grounded-airbus-a380-5955 [12:57] FrankOlken1: Where are we in the communique? [12:59] GaryBergCross: If you want a link to conceptualization you might use this one from the 2007 Summit - http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2007_FrameworksForConsideration#nid10D1 In the context of computer and information sciences, an ontology is a specification of a conceptualization. That is, it specifies the concepts, ideas, relations, abstractions, and so forth in an objective form. The intent is to clarify the meaning, enabling shared understanding. This is the conventional sense of specification in computer science, analogous to the terms requirements specification, database specification, and program specification. In the context of knowledge representation in particular, an ontology specifies the conceptual primitives for representing a domain, in the same way that a database schema specifies the relations used in a database, and a programming language provides the primitives used in an implemented algorithm. [13:02] DougFoxvog: In our terms ontology is far more than "a specification of a conceptualization." It is a FORMAL specification of a conceptualization expressed in a logically formal language. [13:02] CoryCasanave: The precision of ontologies provides for deriving value from models by employing rules, infreence, transformation and analytics. Such automated derivitive value is only possible where the semantics of the models is well understood and represdented in a computational form. [13:07] MikeBennett: An ontology is a formalization of a conceptualization - but what sort of formalization of what sort of conceptualization? the answers to those two questions i.e. what formalism is used (FOL etc.) and what the conceptualization is (relation of model to things modeled; semiotics) tells you what sort or even whether it is an ontology, I would humbly submit. [13:07] GaryBergCross: In Modeling section you have a paragraph that starts with "Computer based modeling languages. " This probably should be hyphenated as "Computer-based modeling languages". [13:10] GaryBergCross: @Doug. As part of the 2007 Summit there was further discussion of conceptualization. Michael Uschold argued that an ontology is not a "specification of an conceptualization" at all, rather, it is a formal way of expressing a conceptualization. [13:18] DougFoxvog: @Gary: "a formal way of expressing" something seems to me to be a "specification" of it. Anyway, i agree with the focus on "formal". [13:25] DougFoxvog: There was a suggestion that we should refer to the PSL ontology when referring to ontologizing events & processes. I urge that this not be done. The PSL ontology doesn't have events & processes & does not allow for subclasses of "activity" See the descriptions of two of the axioms: "Axiom 9 Everything is either an activity, activity occurrence, timepoint, or object." & "Axiom 13 An activity occurrence is associated with a unique activity." [13:30] GaryBergCross: The Use of Ontologies in Requirements Engineering is discussed in an article with that title at ; http://globaljournals.org/GJRE_Volume10/1-The-Use-of-Ontologies-in-Requirements-Engineering.pdf [13:57] Rex Brooks: How about "Big Systems Engineering and Architecture can garner..." [14:00] MikeBennett: There are some good pointers about ontological commitment in the paper referenced by @Gary above. [14:02] DeborahMacPherson: Architecture, engineering, and specifications [14:02] Rex Brooks: Hurrah! [14:03] DeborahMacPherson: arhictecture, engineering, specifications, standards, and modeling -------------