ppy/chat-transcript_unedited_20120119a.txt Chat transcript from room: summit_20120119 2012-01-19 GMT-08:00 [09:20] PeterYim: Welcome to the = OntologySummit2012: Session-02 - Thu 2012-01-19 = Summit Theme: OntologySummit2012: "Ontology for Big Systems" Session Topic: Ontology for Big Systems - What's In Scope? Session Co-chairs: Dr. LeoObrst & Dr. NicolaGuarino Panelists: Track-1: Large-scale systems engineering - Co-Champions: Dr. HensonGraves, Mr. CoryCasanave Track-2: Large-scale engineered systems - Co-Champions: Dr. MatthewWest, Dr. HensonGraves Track-3: Challenge: ontology and big data - Co-Champions: Mr. ErnieLucier, Ms. MaryBrady Track-4: Large-scale domain applications - Co-Champions: Dr. SteveRay, Dr. TrishWhetzel, Mr. CoryCasanave X-Track-A1: Ontology Quality and Large-Scale Systems - Co-champions: Dr. AmandaVizedom Communiqué Co-Lead Editors - Dr. ToddSchneider (in absentia), Mr. AliHashemi Public Relations - champion: Mr. AliHashemi Session page: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2012_01_19 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: == . [08:19] Jack Ring: What slide deck is being used? [08:25] Jack Ring morphed into JackRing [09:14] LeoObrst: I think Peter is still placing the slides. [09:17] PeterYim: slides are up now [09:19] anonymous morphed into TomTinsley [09:25] anonymous morphed into MikeRiben [09:26] Nicola Guarino: Hi everybody [09:27] LeoObrst: Hi, Nicola. [09:28] anonymous morphed into ChristopherSpottiswoode [09:30] JoelBender: getting "please continue to hold while we locate this subscriber" message and music [09:31] anonymous morphed into KenAllgood [09:31] anonymous morphed into dougFoxvog [09:31] AliHashemi: Hi Joel, are you sure you are dialing into the correct room? We're in the call right now. [09:32] JoelBender: redialed - all set [09:32] anonymous morphed into ElizabethF [09:34] anonymous morphed into IAM [09:34] anonymous1 morphed into AndreaWesterinen [09:35] anonymous morphed into MaryBrady [09:38] anonymous morphed into RamGouripeddi [09:39] DuaneNickullisAsleep morphed into DuaneNickull [09:39] DuaneNickull: Good Morning all! [09:41] anonymous morphed into Gary Berg-Cross [09:42] RosarioUceda-Sosa: Good day to everyone. A comment. Given that we all have different backgrounds and interests, it would be a good idea for the track leaders to post a reading list of papers/materials that they think are foundational/relevant/thought provoking. The track leaders could 'filter' the references if the lists become unmanageable. [09:43] PeterYim: @dougFoxvog - thank you for helping post the (chat-room) attendees roster to the wiki session page [09:45] Ram D. Sriram: I will be leaving early, but we need to decide on the date. We have two options: April 12-13 and April 23-24. [09:47] Nicola Guarino: @Rosario: excellent idea [09:48] AliHashemi: +1 to Rosario's idea as well, it would be very helpful. [09:50] RosarioUceda-Sosa: I can't access the slides anymore. Get 'Connection to the server was reset' Anybody else has the same problem? [09:51] PeterYim: @RamSriram - we will let you and MichaelGruninger decide on the symposium date. Just tell us. [09:52] Ram D. Sriram: @Michael: Is April 12-13 fine with you. [09:53] AliHashemi: @Ram, MichaelG is not in the chat. [09:55] PeterYim: @Ram - let's decide at the organizing committee meeting tomorrow, and then announce the decision to the community [09:57] KenAllgood: Could definitely use resources for the group around Automatic Programming if that will remain a focus of Track 3 discussions. [09:57] PeterYim: whoever it is who typed in the box next to the hand, the chat input box is further to the left, please re-type your input, thanks [09:58] anonymous morphed into KenAllgood [10:03] LarryLefkowitz: I was hoping that Track 4 would include -- as part of "large-scale" systems -- those that require large/complex ontologies, not just large data. Is that the case? I didn't see that aspect described in the slides just shown. [10:04] SteveRay: Yes, we hope to cover that as well. That is what was meant by "very complex data sets". [10:06] LarryLefkowitz: @Steve: Great. Let's talk more because I think the complexity of the required ontology is (or can be) distinct from the complexity of the data. That is, I agree that semantic systems need to model the data, but IMHO also need to model the "domain". Either or both of these models can be large/complex. [10:06] SimonSpero: @Steve @Larry - there's also different degrees of "very" (large and/or complex). [10:07] James Odell: There is also a different between complicated and complex. [10:07] RosarioUceda-Sosa: @Steve @Larry -- If that's the case, the comment I made to the 3d track really applies to track 4. Large data sets (instance-level) and large models/metadata (class-level) [10:09] LarryLefkowitz: Wow, and I thought Clinton was a tough English sentence parser in his "what is is" days :-) Yes, I agree with each of those distinctions. [10:09] AndreaWesterinen: There are multiple issues with complex ontologies - both defining (understanding) and storing them, and then reasoning over them. We might want to separate these issues. [10:09] RosarioUceda-Sosa: @Steve @Larry -- Sorry, I hit enter before finishing. If track 4 is explicitly tackling the two dimensions and their different assumptions (storing/streaming/caching data in the instance-level and integrating/prescribing/describing at the class level) it should be a very interesting discussion. [10:10] ChristopherSpottiswoode: @Amanda, Are you distinguishing between ontology quality and ontology-based application quality? [10:10] James Odell: --as well as executing over some part of an ontology (particularly process ontologies) [10:12] AndreaWesterinen: Has anyone looked at the DQM vocabulary for some aspects of the quality work? [10:13] ChristopherSpottiswoode: @Amanda. Thank you! [10:14] SimonSpero: @andrea the RPI LODQ ? [10:14] KenAllgood: Excellent discussion Amanda.. Thanks for offering it up as a track! [10:15] LinePouchard: I looked at some 2005 work by Stuart Madnick at MIT business school. Also an interesting paper "Data Quality- What can ontological ananlysis contribute?" by Andrew Frank. Interesting. Not related to LOD I believe [10:15] Gary Berg-Cross: One might argue that an ontology may have logical quality and/or formal representation quality and/or domain relevance/faithfulness quality...and of course some quality of how these are related in an ontology. [10:17] MikeBennett: Another aspect of ontology quality I think is ientifying what are the appropriate requirements to be met in a business conceptual ontology versus an operational ontology (representing overall business meaning versus decidable ontologies for semantic apps). I don't know if that's a distinction that's widely shared. [10:17] AndreaWesterinen: I do not believe that DQM is related to the LODQ, but they might be. Info on DQM is available at http://semwebquality.org/mediawiki/index.php?title=DQM-Vocabulary_Primer. [10:17] MatthewWest: @Gary: I like those distinctions. [10:17] AndreaWesterinen: @Gary: +1 [10:18] anonymous morphed into RexBrooks [10:19] AndreaWesterinen: @Mike: DQM is related to requirements. [10:20] AliHashemi: (i will be away from the phone for 5 min) [10:20] Gary Berg-Cross: What is the twitter handle for the 2012 Summit again? [10:21] LarryLefkowitz: @Gary: To your list, I'd add an efficiency/utility/effectiveness dimension. Although it might bother the purists who might think that the ontological model should be independent of its use, there is definitely an impact of the design of the ontology and its usefulness. [10:22] ChrisW: This all seems, in contrast to previous years, highly fragmented and unfocused. Will there be a higher level effort to focus the outcome of the summit? [10:22] Nicola Guarino: @MikeBennett: your distinction between ontology quality wrt conceptual understanding and ontology quality wrt operational efficiency is very important, in my opinion [10:22] anonymous morphed into KathyEllis [10:23] BruceBray: Shorter subject titles would be better than longer ones:) [10:23] Gary Berg-Cross: @Larry, I guess I was folding efficiency/utility/effectiveness into the Domain part, but is is worth having these considered as part of pragmatic application of an ontology. [10:24] MatthewWest: @Nicola and @Mike: Indeed this has been a key concern in my area of interest. [10:24] SteveRay: @Chris: Our hope is that we will end up with some specific recommendations about where ontology is particularly well suited to help, be it in the systems engineering process itself, or in the engineered system that results. [10:24] KenAllgood: @Larry: would second your proposition.. [10:25] anonymous morphed into NancyWiegand [10:25] anonymous1 morphed into JosephSimpson [10:25] AmandaVizedom: Capturing here my response to @Christopher on the call, for record: Yes, I am making that distinction, and I am talking about ontology quality specifically. However, the door is open to discuss way in which ontology-based application quality may depend on ontology quality, and how ontology quality in a big systems context can be defined. Additionally, I briefly took off my Champion hat and put on my experienced practitioner hat and stated my view that (applied) ontology quality cannot be evaluated in full without considering (aspects of) the performance of the system of which it is a part. [10:25] LinePouchard: Hi Nancy [10:26] AmandaVizedom: @Gary (and all): twitter hashtag is #ontologysummit2012 [10:26] AliHashemi: (back) @ChrisW - I think more of a focus will emerge as the tracks become more crystallized [10:27] SimonSpero: (it's a question) [10:27] AliHashemi: Peter suggests to post questions here, and use the hand for indicating you want to speak. [10:27] SimonSpero: I asked if long was a dimension of big [10:27] ChrisW: ...in my experience, "focus" is not an emergent quality [10:28] AmandaVizedom: Thanks to everyone who is making comments related to quality, and suggesting references. [10:30] ChristopherSpottiswoode: Thanks Amanda - your answer, now also on the chat, was good. [10:30] SimonSpero: I asked if keeping data usable over archival timespans [10:31] SimonSpero: 50 years and longer [10:31] Gary Berg-Cross: #ontologysummit2012 seems to be the twitter handle. [10:31] MikeBennett: @Nicola and @Matthew glad to hear I'm not alone. We are starting to articulate some of these distinctions in some of our ongoing proof of concept work. It's a bit new to some of the SemTech folks :) [10:31] PeterYim: ref. SteveRay's suggestion (verbal) on prefixing mailing list posts - please use [SystemEngineering], [EngineeredSystems], [BigDataChallenge], [Applications] & [Quality] when discussing respective track issues (I will document that on the wiki too, just in case people want to refresh their minds - it'll be under: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012#nid309T [10:31] AliHashemi1: A preliminary unifying narrative I see is the recommendations for various practitioners. It would seem to begin with identifying what exactly big systems are, and then explicating the various ways that ontologies apply. This would cover tracks 1 & 2. Track 4 supplements those two by providing examples, while track 3 focuses on a problem that persists at the moment. [10:31] ChristopherSpottiswoode: @Amanda, here's another distinction: between functional quality and non-functional quality. [10:32] NancyWiegand: I have an interest in what might be Track 1 or Track 4 regarding how computer architectures hold and use ontologies and other semantic components. For starters, I was thinking of looking at CUAHSI or iPlant to find out how they designed semantics into their architectures. Ilya Zaslavsky and Blazej Bulka from those respectively did answer an email on this and might be willing to give presentations. Others probably know of other systems too. [10:32] SimonSpero: [dataset provenance, citation tracking, replication awareness, etc are important issues in large scale systems engineering) [10:33] NancyWiegand: Line-- Maybe you could talk about your system too, regarding my comment. [10:33] MatthewWest: To try to answer the question being asked about multiple data sets with the same data. What I hope will happen is that the idea of "Authoritative Source" will gain credence, so that those that are the originators of data and are properly the authoritative source will make it available so that the many do not have to cureate their own copy. [10:34] AmandaVizedom: @Larry, your point a way back about "complexity of the required ontology" can also be viewed through the lens of quality and evaluation, in this sense: one significant barrier to effective ontology evaluation is the lack of effective specifications of ontology *requirements*. Understanding how to specify ontology requirements, including what characteristics/dimensions matter to fitness-for-purpose, is one area in which, IMHO, relatively rapid progress in support of ontology quality could be made. [10:35] SimonSpero: It helps [10:35] AliHashemi1: @Chris, the reason I suggested that a focus would emerge is that it is not wholly clear to what extent of the different types of systems will be covered. Once the exact scope of the large-systems being investigated (a function of who is involved in the tracks in addition to the mission statements) will greatly narrow the focus [10:36] SimonSpero: @Chris - I think that the cross-tracks will also help lock in [10:37] anonymous morphed into PavithraKenjige [10:37] Gary Berg-Cross: When we talk about ontology to help Big Data are we talking about Big Meta-Data? Might play well. [10:37] KenAllgood: @Gary: How would you differentiate the two? [10:38] SimonSpero: @Gary "All data is metadata" [10:38] TerryLongstreth: @MatthewWest: your point about "authoritative source" is well taken, but our current models all assume that the authority is a human. In an archiving environment, transferring authority over the course of years is a hard problem. [10:38] PavithraKenjige: What are big systems? how do you define large systems? did anyone discuss it already? [10:38] AliHashemi1 morphed into AliHashemi [10:38] Gary Berg-Cross: Nicila mentioned FuturIcT at last year's Summit, so perhaps an update on that is in order. [10:38] KenAllgood: @Simon: And all metadata is data [10:40] Gary Berg-Cross: @Ken "How would you differentiate the two?" Ontology is formalized metadata??? [10:40] MatthewWest: @Nicola: I agree that an important aspect of the systems we are looking at that is about being intentionally constructed, indeed intentionally construct systems would have been my choice of name for clarity, but for a wider audience I think engineered systems is better, and matches systems engineering. [10:40] LarryLefkowitz: @Amanda: Agreed. Walking this back even one step further, the question might be how ontology development (including requirement specification) should be driven: in the abstract (e.g, "let's build a weather ontology") vs application-driven. In the latter case, of course, one wants to aim for reusability, but there are definite advantages to having a concrete use case as a driver and as a means of validation. [10:41] SimonSpero: @Nicola, @Matthew: Taxis, not Kosmos, in Hayek's terms (Made order, vs. Grown order) [10:41] KenAllgood: @Gary: It could be described as such, I asked the question to gain insight into how it could be presented cleanly.. [10:41] AndreaWesterinen: But, use cases of 1 always work and rarely result in designs that are reusable. [10:41] Gary Berg-Cross: Nicola, thank you for the update. It seems that the project must be considering some of the issues surfaced here. [10:42] AliHashemi: @Nicola and Matthew, I asked on the list, but i'll repeat here - consider someone trying to understand and do urban planning human's role in ecologies / environment climate science are the people trying to understand (and in some case, conduct interventions in these systems) engaged in engineering? Why not _Designed_ Systems ? [10:43] LinePouchard: @Gary: in simulation results on large supercomputers, order of magnitude in size still is one of the differences between data and metadata. Another one maybe formats, with results as scalars or arrays, and metadata as somewhat human-readable. [10:43] AndreaWesterinen: I am not arguing against use cases, just arguing for multiple ones if reusability is a goal. [10:43] Gary Berg-Cross: Design is part of Engineering as is analysis and implementation. [10:44] Nicola Guarino: @Ali: I agree with you, "designed" is better than "engineered". However, I want to include also the context where designed systems are placed in the notion of system. That's why I am proposing "large scale socio-technical systems" [10:45] AliHashemi: @Nicola, that works too, i don't think the notion of function is restricted to engineered systems only. [10:45] Gary Berg-Cross: @Line. That magnitude factor is a great distinction. I was thinking if we are talking about VERY Big data, then we get to Big metadata... [10:45] Nicola Guarino: @Ali: agreed [10:46] LarryLefkowitz: @Andrea: I concur (re multiple use cases). One proxy for that is drawing on the experience (embodied both in terms of an extant ontology and in the minds of the practioners)to design for reuse, even in the case of a single (new) app/usecase. I think we both agree that several are better than one and that one is better than none (i.e., developing in the abstract). [10:46] AliHashemi: @Gary, my point is that someone who is engaged in design planning, or someone who is trying to build a climate model, could use things from engineering, but would not necessarily define their system as an "engineered system" [10:46] AliHashemi: er, could use learning from ontology* [10:46] AliHashemi: design planning --> urban design planning* [10:47] MatthewWest: @Pavithra: yes this has been discussed. We got as far as the whole system being beyond the comprehension of a single person. [10:48] Nicola Guarino: On the ontology of functions, let me point to the recent special issue of Applied Ontology http://iospress.metapress.com/content/u015875jv78t/?p=ccdb1d13b00943ba931aff599352fd21&pi=2 [10:48] Gary Berg-Cross: @Ali. I see your point in an urban design planning activity. The resulting urban product might not be called "engineered", but this seems in part to me one of labeling. [10:50] SimonSpero: Tall for a philosopher, but short for a basketball player... [10:50] AliHashemi: @Gary, I'm also thinking in terms of making the content findable, and such that someone in non-ontology domain would identify with it. If part of the goal of this summit is to foster inter-disciplinary knowledge transfer, then it seems useful to make the terms accurately reflect how different cultures identify their domains. [10:51] SteveRay: @Matthew: But we have included "Large" to also mean simply Big, so I would include Facebook in scope. [10:51] SimonSpero: @MatthewWest was OS/360 a complex system? [10:52] AndreaWesterinen: @Ali: I agree. That was my point earlier about the different aspects of building, understanding, efficiently storing and then reasoning over ontologies. [10:52] AmandaVizedom: @Gary: regarding your suggestion of different types of quality: In the Unscoped Wilds, there is (as you know) a spectrum of views regarding the universality of standards of ontology quality. At the poles of this spectrum are, roughly, (1) those who see ontology quality as context independent (purely formal, specifiable without reference to application or acctivity, and (2) those who see ontology quality as entirely context-specific, locally-definable, and expedient. Most practitioners fall between these poles, but there is a relatively continuous disagreement on the matter. One reason I think that there is hope for this discussion is that we are not in the Unscoped Wilds; we are explicitly looking at ontology quality and Large-scale systems. Letting the systems discussion shape the quality discussion gives us a way to skip the most intractable parts and focus on quality *as it relates to large-scale systems*, that is, as it affects such s! ystems and/or the engineering of them, and as it can be measured and so made useful within these contexts. In this scope, Quality can be operationally defined as ~~ fitness for purpose, where more and less formal criteria can be identified as characteristics contributing (variably) to ontology quality. So, with that in mind, I'd suggest re-casting these distinctions from "kinds of quality" to features relevant to quality, and then asking when and how those features matter, and whether and how we can measure and evaluate them. [10:52] Gary Berg-Cross: Complex in relation to usual human ability to understand...but not for a chess grandmaster... [10:52] AliHashemi: @Andrea, I agree. Especially if track 2 is about ontologies that can be used to describe these systems, they will often include aspects that aren't often within the scope of traditional engineering professions. [10:53] ErnieLucier: RosarioUceda-Sosa requested a reading list. One reference I like is Ultra-Large-Scale Systems: The Software Challenge of the Future 2006, http://www.sei.cmu.edu/library/assets/ULS_Book20062.pdf [10:54] SimonSpero: I agree: If enough people say "it [10:54] SimonSpero: is big system", that's enough for now [10:54] Gary Berg-Cross: @Amanda I do like your related quality to the largeness and conplexity focus we have. [10:54] ChrisW: i suggest it would be a mistake to try and define "complex" and "big" ... even "system". It would take a lot of time, generate something most people would simply disagree with, and serve no real purpose [10:54] JosephSimpson: For complex systems you must first define the boundary of the system. If you can not clearly define the boundary, it is probably is complex. [10:55] ChrisW: rather, what seems to be clear here is a nice focus on ontology quality [10:55] ChrisW: (...looking for focus...) [10:55] AliHashemi: @Chris, would it be useful to identify types of systems that fall within scope of the summit? [10:55] SimonSpero: @ChrisW++ [10:55] Gary Berg-Cross: Are Big systems from 1970 now considered Big in 2012? [10:55] AliHashemi: At least try to make clear which systems we expect our recommendations to apply to, and which do not. [10:55] KenAllgood: @ChrisW: Agree, and we've run into that ourselves already [10:56] AliHashemi: types --> interpretations* [10:57] AmandaVizedom: @Terry: true of some systems, but notably not true of, e.g., sensor fusion systems (which I think would count as Big in several dimensions! [10:58] ChrisW: i think it woudl be useful to make a "big systems may include systems like ..." and make a clearly non-exhaustive list, and then move on [10:59] ChrisW: they key being to move on [10:59] MikeBennett: Apologies, have to drop off for another call. Great session! [10:59] AliHashemi: Also, isn't track 2 answering that question to a large degree? [11:00] KenAllgood: @Chris: Or at the least arrive at a list of criteria. [11:00] LarryLefkowitz: What are the next steps? [11:01] ChrisW: gotta go [11:01] SimonSpero: Would it help to pull a list of the domains that are currently being addressed by the NSF Datanet projects? [11:03] MatthewWest: @Ali: I don't have a fixed idea of what the name of track 2 should be. If someone makes a suggestion and several others support it, I would be inclined to go with it. As far as designed systems is concerned I think it has similar problems of interpretation as engineered systems. You only need to take a narrow view of the name as you have done with engineered systems. [11:03] Gary Berg-Cross: @GChrisW I think the giving examples of Big Systems is a good one.. [11:04] SteveRay: @Gary: That's what Track 4 will try to do. [11:04] MatthewWest: I'm on linked in so I can say something there. [11:05] AliHashemi: There should be some LinkedIn groups where a thread might pick up steam (and could point here) [11:05] SteveRay: Thanks all. [11:05] AliHashemi: Thanks [11:05] Gary Berg-Cross: Bye all. [11:05] SimonSpero: Thanks everyone [11:05] KenAllgood: Thanks Leo and Steve.. [11:05] LeoObrst: Thanks, folks! [11:05] PeterYim: great session ... thanks! [11:06] Nicola Guarino1: bye all! [11:06] PeterYim: -- session ended: 11:05am PST --