ppy/OntologySummit2013_s04_chat-transcript_edited_20130207b.txt ------ Chat transcript from room: summit_20130207 2013-02-07 GMT-08:00 [PST] ------ [08:37] PeterYim: Welcome to the = OntologySummit2013: Virtual Panel Session-04 - Thu 2013-02-07 = Summit Theme: Ontology Evaluation Across the Ontology Lifecycle * Summit Track Title: Track-C: Building Ontologies to Meet Evaluation Criteria Session Topic: Ontology Development Methodologies for Integrating Ontologies * Session Co-chairs: Dr. MatthewWest (Information Junction) & Mr. MikeBennett (EDM Council; Hypercube) Panelists / Briefings: * Professor BarrySmith (University at Buffalo, US) - "Ontological realism as a strategy for integrating ontologies" * Mr. ChrisPartridge (BORO Solutions, UK) - "Ontology Architecture - Top Ontology Architecture" * Mr. AnatolyLevenchuk (TechInvestLab, RU) - "ISO 15926 Reference Data Engineering Methodology" * Mr. MikeBennett (EDM Council; Hypercube, UK) - "Quality Considerations for an Industry Standard Ontology" Logistics: * Refer to details on session page at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_02_07 * (if you haven't already done so) please click on "settings" (top center) and morph from "anonymous" to your RealName (in WikiWord format) * Mute control: *7 to un-mute ... *6 to mute * Can't find Skype Dial pad? ** for Windows Skype users: it's under the "Call" dropdown menu as "Show Dial pad" ** for Linux Skype users: please note that the dial-pad is only available on v4.1 (or later or the earlier Skype versions 2.x,) if the dialpad button is not shown in the call window you need to press the "d" hotkey to enable it. Attendees: AmandaVizedom, AnatolyLevenchuk, AndreyBayda, BarrySmith, BobSmith, BobSchloss, BobbinTeegarden, BruceBray, CarmenChui, ChrisPartridge, DaliaVaranka, DavidMakovoz, DeborahMacPherson, Dmitry, DmitryBorisoglebsky, DougFoxvog, FabianNeuhaus, FranLightsom, GaryBergCross, HansPolzer, hevok, JaguaraciSilva, JeffersonBraswell, JoanneLuciano, JoaoPauloAlmeida, Joe, JoelBender, JohnBilmanis, LamarHenderson, LeoObrst, LowellVizenor, MaksK, MarcelaVegetti, MariCarmenSuarezFigueroa, MatthewWest, MeganKatsumi, MichaelGruninger, MichaelRiben, MikeBennett, MikeDean, MikeDenny, NikolayBorgest, PeterYim, RichardMartin, RobHausam, RobertRovetto, ScottHills, SimonSpero, SteveRay, TerryLongstreth, TillMossakowski, ToddSchneider, TorstenHahmann, TrishWhetzel, VictorAgroskin. == Proceedings: == [09:04] anonymous morphed into Joe [09:14] anonymous morphed into CarmenChui [09:25] anonymous1 morphed into TorstenHahmann [09:27] anonymous1 morphed into RobertRovetto [09:29] anonymous1 morphed into RobHausam [09:30] anonymous1 morphed into MichaelRiben [09:30] anonymous1 morphed into TrishWhetzel [09:31] anonymous1 morphed into DougFoxvog [09:31] anonymous morphed into NikolayBorgest [09:32] anonymous morphed into MikeDenny [09:32] Andrey Bayda morphed into AndreyBayda [09:34] anonymous morphed into DmitryBorisoglebsky [09:34] anonymous1 morphed into ToddSchneider [09:34] anonymous morphed into BarrySmith [09:34] anonymous2 morphed into HansPolzer [09:34] anonymous morphed into ChrisPartridge [09:34] anonymous1 morphed into JoanneLuciano [09:37] PeterYim: == MatthewWest opens the session on behalf of the co-chairs ... see: the [0-Chair] slides [09:37] anonymous morphed into hevok [09:40] anonymous1 morphed into LamarHenderson [09:41] anonymous morphed into JeffersonBraswell [09:43] List of members: AmandaVizedom, AnatolyLevenchuk, AndreyBayda, BarrySmith, BobSmith, BobbinTeegarden, CarmenChui, ChrisPartridge, DaliaVaranka, DmitryBorisoglebsky, DougFoxvog, FabianNeuhaus, FranLightsom, HansPolzer, hevok, JoanneLuciano, JoaoPauloAlmeida, JeffersonBraswell, Joe, JoelBender, JohnBilmanis, marcelaVegetti, MariCarmenSuarezFigueroa, MatthewWest, MeganKatsumi, MichaelGruninger, MichaelRiben, MikeBennett, MikeDean, MikeDenny, NikolayBorgest, PeterYim, RichardMartin, RobertRovetto, RobHausam, ScottHills, SteveRay, TerryLongstreth, TillMossakowski, ToddSchneider, TorstenHahmann, TrishWhetzel, VictorAgroskin, vnc2 [09:41] PeterYim: == BarrySmith presenting ... see: the [1-Smith] slides [09:41] anonymous morphed into TrishWhetzel [09:46] LeoObrst: Hi, folks, running late. [09:46] anonymous morphed into Dmitry [09:46] BobSchloss: BarrySmith - for the future, your slide 4 meant to say XML but says XLM :-) [09:57] PeterYim: @BobSchloss, @BarrySmith - I've updated the slides (and it says "XML" on Barry's slide#4 now) ... Thank you, Bob, for the prompt. [09:49] MatthewWest: [ref. BarrySmith's presentation showing multiple hits of obesity when that term is searched in BioPortal] Whose definition of obesity should everyone else use? [09:56] HansPolzer: Another way to constrain ontology for purposes of integration is to be explicit about the specific ontology being used by the participating entities [09:58] HansPolzer: I would suggest that explicitness is often better than "realism". Explicitness entails being explicit about the levels of realism in Barry's presentation [10:03] BarrySmith: HansPolzer writes "Another way to constrain ontology for purposes of integration is to be explicit about the specific ontology being used by the participating entities" -- this would indeed have some positive effect, but it is not clear how it would serve to constrain; on the contrary, it seems that it might well encourage further proliferation [09:57] AmandaVizedom: Assume for the moment that we could somehow enforce such uniform terminology, despite the long history of failed attempts to do so. How, then, do you address the real cognitive performance issues of forcing users (human reasoners, information consumers) into vocabulary and information processes not native to their expert fields, not adapting and growing with local dynamics? Cognitive Science says this will harm their performance, especially in high-stakes, high-uncertainty, time-sensitive fields. How do you suggest mitigating this harm, and why is this terminology-focus worth it? [10:03] BarrySmith: AmandaVizedom asks how the realist would address the real cognitive performance issues of forcing users (human reasoners, information consumers) into vocabulary and information processes not native to their expert fields not adapting and growing with local dynamics? Cognitive Science says this will harm their performance, especially in high-stakes, high-uncertainty, time-sensitive fields. How do you suggest mitigating this harm, and why is this terminology-focus worth it? [10:05] BarrySmith: AmandaVizedom asks how the realist would address the real cognitive performance issues of forcing users (human reasoners, information consumers) into vocabulary and information processes not native to their expert fields not adapting and growing with local dynamics? The answer is that, from the realist perspective, a small fraction of people in any given field would be involved in ontology development, and they would understand the need to use a common vocabulary. Not every disciplinary subdialect needs to be represented in the ontology; that way chaos lies [09:58] JoaoPauloAlmeida: What about if you want to describe social reality? [09:59] JoaoPauloAlmeida: Not part of science text book [10:01] MikeBennett: @JoaoPauloAlmeida what about John Searle's ontology of social constructs? A text book. [10:04] JoaoPauloAlmeida: @MikeBennett All BFO usage examples are from biology, chemistry, ... I was wondering whether Barry thinks it can be applied to social domains. [10:01] JoaoPauloAlmeida: Will that not require a revision of BFO to include "doctrine", ... norms, agents, etc.? [10:06] BarrySmith: JoaoPauloAlmeida asks whether BFO usage can be applied to social domains. We are working on this. See e.g. http://militaryontology.org [10:06] JoaoPauloAlmeida: Thanks [10:02] HansPolzer: social reality is grounded in near term or current social opinion within a scoped population. Be explicit about the scope of that population and you can obtain social reality by polling that population subset. [09:59] HansPolzer: Appeals to authority or standardization to promote integration have limited scalability in scope. [10:00] PeterYim: @Barry - is there a plan/timeline to get all ontologies in the OBO Foundry to be BFO "compliant" (if they aren't already)? [10:01] ToddSchneider: Have to go. [10:01] AmandaVizedom: @BarrySmith: You slide between talk of "common ontology" and talk of controlled terminology. Why? Why not map multiple terminologies (including multilingual) to common ontology and use localization and user modeling techniques? [10:03] DougFoxvog: @Amanda: +1 [09:58] PeterYim: == ChrisPartridge presenting ... see: the [2-Partridge] slides [10:02] anonymous1 morphed into LowellVizenor [10:04] HansPolzer: Re Chris's talk, a look at the NCOIC SCOPE model might be of interest regarding the relationship between different scales of projects/systems/enterprises and architecture. [10:06] JoaoPauloAlmeida: There is a more general definition of architecture that is used by IEEE (1471-2000): The fundamental organization of a system, embodied in its components, their relationships to each other and the environment, and the principles governing its design and evolution. [10:06] MikeBennett: @JoaoPauloAlmeida good question. In FIBO we make reference to JohnSowa's KR Lattice which has similar upper ontology partitions, and have added Social Constructs, but I almost wonder if they should be a partition in their own right. [10:07] PeterYim: @Joe, @hevok, @Dmitry, @anonymous - kindly morph into your real name (click on the "settings" button at the top center of the window to do so) so we all know who's here and properly attribute contributions ... thanks. [10:07] HansPolzer: Shared understanding and "common" ontology raise the issue of what that understanding is shared across and what the ontology is common to. Commonality and sharing imply a scope across which things are shared or common. We need a way to define and share that scope among the participants that are sharing or using a common artifact. [10:08] GaryBergCross: Generative Entrenchment sounds similar to ideas around contingency. [10:10] HansPolzer: Managing a project isn't just about managing interdependence of the parts - although very important, but also managing the relationship of the project/enterprise with the larger ecosystem. Often this latter aspect is assumed and assumed to be static - which it rarely really is! [10:14] HansPolzer: I would add to the Pierce quote that people also have a context assumption and an assumption concerning the scope of that context. [10:15] HansPolzer: This is the source of silos, by the way [10:18] HansPolzer: It would be helpful if people on projects were asked to explicitly identify "adjacent" domains and interacting external projects/systems/enterprises/contexts that their creation will need to interact with. In other words, explicitly consider the ecosystem within which they operate [10:21] BarrySmith: To HansPolzer's remark -- that people identify neighboring domains and build ontologies in such a way as to be consistent with neighboring ontologies -- this is one of the principles of the OBO Foundry: http://obofoundry.org [10:21] MikeBennett: @Hans good point. A related point is to what extent you define abstractions in a given domain ontology, such that terms in an adjacent subject matter may be specializations of those same abstractions. This makes no sense in the application domain but helps with common meaning. [10:26] ChrisPartridge: @HansPolzer - Yes I agree that one needs to look at the larger ecosystem. However, I think you need to differentiate between the ontology as artefact and its ecosystem - and the ontology as the 'reality' being described and its ecosystem. They are usually different, sometimes very different. [10:28] HansPolzer: Chris, agree that these are best managed separately - but linked [10:31] HansPolzer: Chris, this is also similar to the "different path" effect you referenced in your talk. If you look more explicitly at the ecosystem and its dynamic trajectory you may end up with a different solution than if you just look at the problem that the ontology is addressing with only implicit context assumptions. [10:29] JaguaraciSilva: how can we define the difference among artifacts and ontologies if the proper ontology can be an artefact within a ecosystem? [10:35] HansPolzer: Jaguariaci, ontologies are both artifacts and used to describe/represent artifacts. We don't have an ontology of ecosystems and their scope, but we should. Then you could specify the scope of the ontology and the ecosystem assumptions of that ontology. The NCOIC SCOPE model is not an ontology, but it is an effort to provide a descriptive framework for characterizing scope of an artifact/institution/system/ecosystem [10:38] JaguaraciSilva: @Hans, thanks!. it means a "architecture view" in an well defined context. [10:16] anonymous1 morphed into DeborahMacPherson [10:17] anonymous1 morphed into JaguaraciSilva [10:20] MichaelGruninger: @ChrisPartridge: Can you identify an upper ontology for each of the choices on slide 11? Additionally, have you evaluated existing upper ontologies with respect to the meta-ontological choices? [10:31] ChrisPartridge: @MichaelGruninger Hi Mike. I'm not sure one can have an ontology component for each of the choices, as they are tightly coupled. One can take a top ontology and classify which of the choices it has made - and see the outcome. My focus has been on the choices that have been made in the top ontologies I have worked on - and the results of the choices. However, I have commented on SUMO, DOLCE and BFO as I have been exposed to them. My proposal is that the developers of top ontologies should classify them - if not by the choices I propose then by some of their own making - and provide arguments for their choices. [10:46] RobertRovetto: @ChrisPartridge When you say "top ontologies should classify them", what is "them" referring to? The choices made? [10:52] MatthewWest: @RobertRovetto: Yes, he means classify them by the ontological commitments made as e.g. listed on one of his slides. [10:24] PeterYim: == AnatolyLevenchuk presenting ... see: the [3-Levenchuk] slides [10:28] DaliaVaranka1 morphed into DaliaVaranka [10:37] anonymous1 morphed into SimonSpero [10:39] AmandaVizedom: @AnatolyLevenchuk: Can you explain what you mean by "same domain" (on your slide 11)? [11:00] AnatolyLevenchuk: @AmandaVizedom: when we engineer formal symbolic system as artifact that represent something in real world, it is the same activity. Especially if you compare declarative programming with programming, modeling with programming (Simula 68 is a memory about times when programming and modeling was the same), ontologizing and data modeling, etc., you see multiple generalities in essence of this activities but completely different terminology, conferences and even theory. Now it slowly converges (e.g. Domain-Driven Design in programming is close to ontologising, Model-Driven Programming is part of software engineering now, etc.). [11:31] AmandaVizedom: @Anatoly, thanks for your answer. [10:41] HansPolzer: Ref Anatoly's talk, the programming, ontology, and modeling are not the same domain. They are overlapping domains that share many scope dimensions, but differ from each other in specific other scope dimensions (like what they model/represent). It helps to be specific/explicit about along what dimensions they differ so that we can better identify their commonality and the process elements that are appropriate to them and which process elements need to differ and how. [10:55] JaguaraciSilva: by conceptual modeling the view programming, ontology and modeling aren't in the same domain, but if there's another need what characterizes a [system architecture by example] it can result on unique view, what depends on such concerns. [11:06] AnatolyLevenchuk: @HansPolzer: if you want to see differences, you definitely will find them. I want to see commonality to heavy reuse achievements of this professional silos, then I find that all these domain not so distinct in essence of their intent: to engineer an executable (interpretable) formal systems that reflect real world systems. [11:09] HansPolzer: Anatoly, my point is that if you want commonality, that commonality has to deal with the differences across which you want commonality. Trying to force fit commonality in places where there are essential differences (as seen by the domain stakeholders and their purposes) results in empty standards, i.e., ones that aren't followed. [11:15] AnatolyLevenchuk: @Hans: if I tell that tigers and lions are mammals, that is not I will miss striped skin of a tiger and attribute it to lion. But I will feed them with milk early in their life cycle and with meat later. This is my approach for programming, ontologizing, modeling. [11:21] HansPolzer: Anatoly - I understand what you were trying to communicate - I was just pointing out that we have different names for these domains for a reason. In many contexts, these domains may well be indistinguishable - but be sure that they are when you are applying a process or ontology standard to them in that context. [11:04] JaguaraciSilva: @AnatolyLevenchuk do you know some studies with domain-driven design? I've used MDA (Model Driven Architecture) approaches on last years. [11:10] AnatolyLevenchuk: @Jaguaraci: 30 years ago we discuss domain-driven design as "if you not knowing what exactly should do your system, better use bottom up process and build library that reflect your project". Now this is DDD (Domain-Driven Design). Yes, I regularly read about DDD and actively use it. MDA is about different thing (but you can use both). [10:48] PeterYim: == MikeBennett presenting ... see: the [4-Bennett] slides [10:52] anonymous1 morphed into SimonSpero [10:58] HansPolzer: Ref Mike's talk: good point on no right answer for genius versus methodology balance. That's where context and scope come in [11:00] AmandaVizedom: @MikeBennett: Genius-Methodology Balance is an interesting suggestion. I wonder, though, whether there is some choice as to whether this is treated as zero or positive sum. As I think over the places I've worked, those couple that had the highest concentration of really brilliant people *also* paid the most attention to methodology and related areas such as training and testing. This did, however, require additional investment in really brilliant people to lead and coordinate those latter activities! [11:03] anonymous1 morphed into MaksK [11:04] PeterYim: @MaksK - would you be kind enough to morph into your real name, please [11:08] AmandaVizedom: @MikeBennett: regarding "the Bonus" (your slide 12): Another, potentially major bonus is "implementability" of the standard itself -- that is, the usability of the standard in compliance monitoring. Have you seen work in this direction with FIBO? [11:11] SimonSpero: @MikeBennett: A lot of common things that you have to use restriction classes for are easily expressed using things like controlled natural language. e.g. Everything that a carnivore eats is an animal. [11:13] TerryLongstreth: @Simon: And CNL (controlled natural language) can express ambiguity (where resolving ambiguity is an implementation detail); e.g. Almost Everything that a carnivore eats is an animal. [11:14] SimonSpero: @TerryLongstreth: Most is a nasty nasty quantifier :-) [11:11] DougFoxvog: @MikeBennett: A requirement for a *LANGUAGE* to be DL-safe seems to be self-defeating. Businesses regularly use programming languages (!) none of which are DL-safe. The issue always is *HOW* the language is used. Restricting the power of the language is not, imho, an appropriate answer. [11:13] SimonSpero: @MikeBennett: I usually have to check with Attempto to see what it was I just said. [11:15] HansPolzer: @MikeBennett, ref slide 15 - need to be explicit about scope/context of the business domain at issue, as well as any other domains the selected domain needs to interact with and to what extent it needs to do so [11:18] PeterYim: == Q&A and Open Discussion ... [11:18] AmandaVizedom: Question for All Panelists: As you understand & practice it, what role(s) does ontology evaluation play in development methodology. [11:20] AmandaVizedom: (and, to the extent that there are multiple answers, what types of evaluation play those roles)? [11:21] JoanneLuciano: And adding to Amanda's question, I am wondering what thoughts have been given to incremental development and modularization of ontology development and evaluation [11:23] LeoObrst: @Joanne: yes, modularity issues are very important! Enables relatively independent, parallel ontology development. [11:30] JoanneLuciano: @LeoObrst --that's the idea... [11:30] MikeBennett: Modularity has been an important consideration in the development of FIBO. Partly this is so that one can take a specific sub-set of those ontologies, either as application-ready ontologies (ambitious!) or as the basis from which to then derive the operational ontology. [11:32] JoanneLuciano: @MikeBennett -- good. [11:32] ChrisPartridge: @Leo - I have concerns about this idea. On the one hand it is good, but on the other hand complex (i.e. functionally rich) systems are normally tightly coupled. I have a feeling that this desire for modularity could be a kind of self-hard - deliberately dumbing down the system. [11:34] LeoObrst: @Chris: yes, there are serious issues. You must do an analysis (top-down) initially to characterize the modules and their dependencies. [11:35] LeoObrst: @Chris: more later on this topic! [11:34] JoanneLuciano: We haven't talked much about the relationship of OWL and RDF... [11:34] JoanneLuciano: and multiple ontologies over the same domain [11:34] JoanneLuciano: (another time!) [11:23] SimonSpero: Further to AmandaVizedom's question: in agile methods, continuous tests are generally considered critical (Unit -> Integration -> Behavior). [11:23] SimonSpero: Can there be Onto-Unit (without Application) [11:25] HansPolzer: Continuous evaluation throughout development might be practical for evaluating attributes that can be analyzed via automated methods, but unlikely to be practical if it requires lots of human stakeholders and expertise and time. A more phased approach is probably more pragmatic in most contexts, with maybe some intrinsic attributes evaluated more frequently during the development process [11:28] JaguaraciSilva: @Hans: continuous evaluation can using a continuous integration environment such as Hudson, TFS, etc.. [11:29] HansPolzer: @Jaguaraci: Yes [11:26] SteveRay: [ref. ChrisPartridge's verbal remarks about "mentoring is more important" - citing: "give people some food and they will be starving tomorrow; give them a line and a hook, and they will be able to eat for the rest of their lives"] That was Lord Kelvin who said that. [11:27] ChrisPartridge: @Steve - Yup, thanks. I've had a long day. [11:28] AmandaVizedom: [ref. verbal remarks] Thanks for your answers, @Chris and @Mike! [11:30] MatthewWest: @Barry - *7 to unmute [11:30] ChrisPartridge: @Barry - is there a BarrySmith2 - if so, is this a good thing? [11:31] PeterYim: @Matthew & Fabian - Barry *is* on the voice line now [... BarrySmith's verbal remarks followed.] [11:34] PeterYim: great session! [11:34] LeoObrst: Thanks, all! [11:34] PeterYim: join us again, same time next week, for OntologySummit2013 session-05: "Software Environments for Evaluating Ontologies - I " - Co-chairs: MichaelDenny (MITRE) & PeterYim (Ontolog; CIM3) - http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_02_14 [11:35] PeterYim: -- session ended: 11:34 am PST -- ------