Actions

Ontolog Forum

Ontology Summit 2013: Panel Session-06 - Thu 2013-02-21

Summit Theme: "Ontology Evaluation Across the Ontology Lifecycle"

  • Summit General Co-chairs: Professor MichaelGruninger (U of Toronto, Canada) and Dr. MatthewWest (Information Junction, UK)

Session Topic: Ontology Summit 2013: Synthesis-I

  • Session Chair: Dr. MatthewWest (Information Junction, UK) - intro slides

Briefings:

  • Dr. MatthewWest (Information Junction, UK) & Professor MichaelGruninger (U of Toronto, Canada) - "General Assessment & Fine-tuning of Ontology Summit 2013 Direction & Approach" [ slides]
  • Dr. LeoObrst (MITRE) & Dr. SteveRay (CMU) - "Track-A: Intrinsic Aspects of Ontology Evaluation - Synthesis-1" slides
  • Mr. TerryLongstreth (Ind. Consultant) & Dr. ToddSchneider (Raytheon) - "Track-B: Extrinsic Aspects of Ontology Evaluation - Synthesis-1" slides
  • Dr. MatthewWest (Information Junction) & Mr. MikeBennett (EDM Council; Hypercube) - "Track-C: Building Ontologies to Meet Evaluation Criteria - Synthesis-1" slides
  • Dr. MichaelDenny (MITRE) & Mr. PeterYim (Ontolog; CIM3) - "Track-D: Software Environments for Evaluating Ontologies - Synthesis-1" slides
  • Dr. AmandaVizedom (Ind. Consultant) & Dr. FabianNeuhaus (NIST) - "Approach to the Ontology Summit 2013 Communique" slides
  • Mr. MikeDean (Raytheon BBN) & Mr. PeterYim (Ontolog; CIM3) - "Approach to the "Hackathon" & "Clinics" Activities" slides

Archives

Abstract

OntologySummit2013 Session-06: "Synthesis-I" - intro slides

This is our 8th Ontology Summit, a joint initiative by NIST, Ontolog, NCOR, NCBO, IAOA & NCO_NITRD with the support of our co-sponsors. The theme adopted for this Ontology Summit is: "Ontology Evaluation Across the Ontology Lifecycle."

This is our 6th event, and we are a third of the way through the virtual sessions for the Summit. Each of the four tracks have hosted very exciting presentations that address the key Summit themes -- Intrinsic Aspects of Ontology Evaluation, Extrinsic Aspects of Ontology Evaluation, Building Ontologies to Meet Evaluation Criteria, and Software Environments for Evaluating Ontologies.

Currently, there is no agreed methodology for development of ontologies, and there are no universally agreed metrics for ontology evaluation. At the same time, everybody agrees that there are a lot of badly engineered ontologies out there, thus people use -- at least implicitly -- some criteria for the evaluation of ontologies.

During this Ontology Summit, we seek to identify best practices for ontology development and evaluation. We will consider the entire lifecycle of an ontology -- from requirements gathering and analysis, through to design and implementation. In this endeavor, the Summit will seek collaboration with the software engineering and knowledge acquisition communities. Research in these fields has led to several mature models for the software lifecycle and the design of knowledge-based systems, and we expect that fruitful interaction among all participants will lead to a consensus for a methodology within ontological engineering. Following earlier Ontology Summit practice, the synthesized results of this season's discourse will be published as a Communiqué.

In today's session, we will focus on the synthesis of all of these ideas as input into an initial draft of the Summit Communiqué. We will also have a discussion on some new features of this year's Symposium -- a hackathon for ontology evaluation software support and the idea of ontology evaluation clinics for ontologies that people want evaluated.

More details about this Ontology Summit is available at: OntologySummit2013 (homepage for this summit)

Agenda

OntologySummit2013 - Panel Session-06

  • Session Format: this is a virtual session conducted over an augmented conference call
  • 1. Opening by the chair (MatthewWest) [5 min.] ... [ slides ]
  • 2. General assessment on how things are developing and fine tuning of direction/approach (MatthewWest and MichaelGruninger) [5 min.]
  • 3. Track Synthesis I (presentation of the interim deliverables by each pair of track co-champions) [10 min/track]
    • 3.1 Track A: Synthesis-1 (LeoObrst and SteveRay)
    • 3.2 Track B: Synthesis-1 (TerryLongstreth and ToddSchneider)
    • 3.3 Track C: Synthesis-1 (MatthewWest and MikeBennett)
    • 3.4 Track D: Synthesis-1 (MichaelDenny and PeterYim)
    • 3.5 Q&A and discussion on the tracks syntheses [~10 min.] ... please refer to process above
  • 4. Approach to the Communique and a proposed Communique Outline (AmandaVizedom and FabianNeuhaus) [15 min.]
    • 4.1 Q&A and discussion on this topic [~10 min.]
  • 5. Approach to the "hackathon" & "clinics" (MikeDean and PeterYim) [15 min.]
    • 5.1 Q&A and discussion on this topic, including brainstorming on possible "hackathon/clinic" projects [~10 min.]
  • 6. Summary/wrap-up/announcements [5 min.]

Proceedings

Please refer to the above

IM Chat Transcript captured during the session

see raw transcript here.

(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.

-- begin in-session chat-transcript --


Chat transcript from room: summit_20130221

2013-02-21 GMT-08:00 [PST]


[8:36] Peter P. Yim: Welcome to the

Ontology Summit 2013: Panel Session-06 - Thu 2013-02-21

Summit Theme: Ontology Evaluation Across the Ontology Lifecycle

Session Topic: Ontology Summit 2013: Synthesis-I

Panelists / Briefings:

- "General Assessment & Fine-tuning of Ontology Summit 2013 Direction & Approach"

- "Track-A: Intrinsic Aspects of Ontology Evaluation - Synthesis-1"

- "Track-B: Extrinsic Aspects of Ontology Evaluation - Synthesis-1"

- "Track-C: Building Ontologies to Meet Evaluation Criteria - Synthesis-1"

- "Track-D: Software Environments for Evaluating Ontologies - Synthesis-1"

- "Approach to the Ontology Summit 2013 Communique"

- "Approach to the "Hackathon" & "Clinics" Activities"

Logistics:

  • (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: Ali Hashemi, Amanda Vizedom, Anatoly Levenchuk, AstridDuqueRamos, Bruce Bray, Clare Paul,

David Leal, Doug Foxvog, Duane Nickull, Fabian Neuhaus, Frank Loebe, GaryBergCross, Jeanne Holm, Jie Zheng,

Joanne Luciano, Joel Bender, Ken Baclawski, Lamar Henderson, Leo Obrst, Marcela Vegetti, Maria Poveda,

Matthew West, Michael Denny, Mike Bennett, Mike Dean, Mike Riben, Peter P. Yim, Richard Martin, Simon Spero,

Steve Ray, Terry Longstreth, Till Mossakowski, Todd Schneider, Tom Tinsley, Yuriy Milov

Proceedings:

[9:25] anonymous morphed into Maria Poveda

[9:27] Steve Ray: Hi Maria, glad you could join us today.

[9:28] anonymous morphed into Michael Denny

[9:29] Maria Poveda: Hi all

[9:30] anonymous morphed into Lamar Henderson

[9:30] Steve Ray: Skype is acting up again. Drops me after about 2 seconds...

[9:31] Steve Ray: Google Voice works...

[9:42] Joanne Luciano: can't get in on skype :-(

[9:45] Peter P. Yim: @Joanne - please try restarting skype (or restarting your machine) or call one of

the phone numbers ... skype should be working as quite a few (over 15) are connected via skype as we speak

[9:32] Duane Nickull: Good Morning all!

[9:33] Anatoly Levenchuk: To Duane: we have 21:13 here in Moscow. Good night! :-)

[9:35] Peter P. Yim: == Matthew West opens the session on behalf of the General Co-chairs

... see: the [0-Chair] slides

[9:35] List of members: Ali Hashemi, Amanda Vizedom, Anatoly Levenchuk, Astrid, Clare Paul, David Leal,

Doug Foxvog, Duane Nickull, Fabian Neuhaus, Frank Loebe, Joel Bender, Leo Obrst, Maria Poveda, Matthew West,

Michael Denny, Mike Bennett, Mike Dean, Peter P. Yim, Richard Martin, Simon Spero, Steve Ray, Terry Longstreth,

Todd Schneider, vnc2

[9:39] Amanda Vizedom: re: Matthew's slide 4: ...or, it may be conscious knowledge on the part of an

individual reviewer, but it isn't shared knowledge. Thus, issues of consistency and guidance for the

field, etc.

[9:42] Doug Foxvog: @Amanda: the ref. to "Matthew's slide 4" should be to Steve/Leo's slide 4.

[9:48] Amanda Vizedom: @Doug: No, in that comment I was responding to Matthew's comment about how we

  • do* evaluation (of ontology papers).

[9:35] Simon Spero morphed into Simon Spero

[9:44] anonymous morphed into Tom Tinsley

[9:39] Peter P. Yim: == Leo Obrst / Steve Ray presenting ... see: the [1-Track-A] slides

[9:41] Amanda Vizedom: re: Leo's slide 2: In-Between > Both

[9:44] Fabian Neuhaus: @Leo: slide 2: I think there is an important difference between relationship

between ontology and world (e.g, whether the ontology represents reality accurately) and whether the

ontology meets black box requirements of an IT system. The first can be evaluated independently of

requirements, the second is always relative to the requirements from an IT system. In the first case

the ontology is not opaque to the tester, in the second it isn't [correction: "it might be" (ref.

below)]. Thus, these should not be lumped together as "extrinsic"

[9:58] Steve Ray: @Fabian, your first comment at 9:44 you had two negatives. Could you restate? (The

sentence beginning "In the first case...")

[9:59] Fabian Neuhaus: Sorry. I meant to write: "In the first case the ontology is not opaque to the

tester, in the second it might be (as blackbox testing). Thus, these should not be lumped together

as "extrinsic"

[10:00] Leo Obrst: @Fabian: (slide 2): Yes, indeed. I don't think we lump these together as

extrinsic, if you look at the other slides. The first focuses on mostly intrinsic-->land of

in-between. The second focuses on the extrinsic regions.

[10:01] Joanne Luciano: @LeoObrst --> I would say land of in-betweenS (plural)

[10:02] Steve Ray: OK. As Leo said, he and I are using "Intrinsic" and "Extrinsic" as useful extreme

concepts, and few if any evaluations will be at either extreme.

[9:51] Amanda Vizedom: @Fabian: I think that there are aspects of the first that can be evaluated

independently, but not nearly enough to select/eval ontologies for most uses -- they don't just need

to represent the world, but the parts and aspects of the world with which the domain / users

interact.

[9:52] Fabian Neuhaus: @Amanda. I agree. However, ontology evaluation is not only done for the

purpose of choice. It is, for example, done during the development process.

[9:53] Michael Denny: @FabianNeuhaus +1 Fitness for an application versus fitness as conformance to

world reality.

[9:54] Fabian Neuhaus: @MichaelDenny: Exactly!

[9:49] Mike Bennett: I think Fabian's point has interesting implications for the creation of formal

methodologies for ontology development and evaluation - in particular the ontology-world

relationship should be fundamental to what process paths to follow in such a methodology.

[9:57] Amanda Vizedom: @Fabian, Michael: I agree that such elements of "conformance to world reality"

can be independently assessed. I argue, however, that in order to evaluate "conformance to world

reality" usefully, you need to know what portion of world is supposed to be modeled. As with

scientific theories, ontologies cover not only what we recognize at large scale as domains, but

particular kinds of relationships and interactions, and the characteristics of things that partake

of those.

[9:58] Mike Bennett: @Amanda +1 - there's the basic model theoretic relation of whether it's really

an ontology of the world or some application, and for the former, there's the scope and the

ontological commitments that would be appropriate to that scope.

[9:59] Joanne Luciano: +100 (have to know purpose before can evaluate)

[9:45] Amanda Vizedom: @Leo: would you put some aspects of reasoning support in Region 1 (Intrinsic)?

I think I would, even though performing the reasoning requires more than the ontology. Probably

somewhat about the language and somewhat about what content is actually represented. I'm thinking

about questions like: is there support for representation of (& reasoning about) uncertainty? Is

there support for provenance information capture, such that it, too can be reasoned about?

[10:02] Leo Obrst: @Amanda: yes, real reasoning figures across the 3 regions.

[9:47] Amanda Vizedom: @Leo: I would also see adequacy of coverage as Region 2, insofar as you can't

tell what aspects of the world are wanted without knowing about the domain / usage.

[10:04] Leo Obrst: @Amanda: (re. your second point) Yes, adequacy of coverage primarily falls under

Region 2, where domain knowledge and ontology-world correspondence is very important.

[9:46] Todd Schneider: Leo, Steve, Why are 'Transitivity, symmetry, reflexivity, equivalence' listed

as meta-properties [to an ontology]?

[9:48] Doug Foxvog: @Todd: "transitivity, symmetry, & reflexivity" are properties of properties; thus

meta-properties. Equivalence can be a property of properties as well as one of types and individuals.

[9:48] Todd Schneider: Doug, okay.

[9:48] anonymous morphed into GaryBergCross

[9:51] Doug Foxvog: Region 3 (purely extrinsic) would disallow a query as to whether two classes are

disjoint. However, since it allows queries, couldn't it ask if a hypothetical thing (perhaps by

reifying it) could be an instance of those two classes -- in order to determine disjointness?

[10:05] Fabian Neuhaus: @Leo: I just don't think that this is a helpful way to slice up the cake. The

way you describe it there is a sliding scale between two extremes with considering only internal

properties on one side and considering behavior on the other side. Evaluating the ontology on

whether the ontology describes reality properly is not "in the middle between the extremes" it is a

different thing entirely.

[10:08] Steve Ray: @Fabian: I don't agree - this is one of the performance characteristics a user

would want to assure is met.

[10:07] Todd Schneider: All, from an IT perspective use of the term 'reality' to describe intended

interpretations or uses (of the IT system) is misleading.

[10:11] Leo Obrst: @Fabian: well, we discussed this and felt that by providing pole perspectives,

that this would help. One would probably say that conformance of the ontology to reality is truly

something that spans all 3 regions, since by definition that is what an ontology as an engineering

construct is all about. One might gauge that in different ways. For example, if one has 2 predicates

or 2 classes and 1 property for an ontology that is supposed to represent a complex domain, one

might gauge it from a narrow intrinsic perspective as being insufficient. That of course is a simple

case one hardly ever finds.

[10:14] Amanda Vizedom: @Fabian, @Leo: I would agree that it doesn't feel like a scale. The second

kind of evaluation isn't really "between" the other two. Rather, it's a different kind of activity

that incorporates some of the kinds of knowledge that go into the first and third (i.e., ontology

theory and understanding of the intended use, respectively), plus some other kinds. To some extent,

you could view this in a Venn diagram manner, where each area represent the kinds of knowledge and

tools you need in order to do one of these three types of evaluation. In that case, 1 and 3 would

intersect, and 2 would include, but not be limited to, that intersection. It might also include

knowledge of human factors research, experimental techniques, and various ways of achieving SME

validation, etc.

[10:16] Fabian Neuhaus: @Amanda: I agree. You can easily build an ontology that scores well on

intrinsic metrics and shows the intended behavior in the sense of answering queries fine, but

contains factual errors.

[10:11] Frank Loebe: @Fabian: Are you aware of any methodology / approach for evaluating an ontology

regarding its performance of describing reality? Or was anything discussed in the previous summit

sessions (not all of which I could follow / catch up with)?

[10:13] Fabian Neuhaus: @Frank: Werner Ceusters addresses that to a degree in his methodology.

[10:08] Fabian Neuhaus: @Todd: that's not what I mean by reality.

[10:09] Todd Schneider: Fabian, so much for interpretation clarity:) But I was also referring to

Leo's & Steve's slides and in general.

[10:28] Till Mossakowski: a question about the track A talk (slide 7): is it possible to download

[[ConferenceCall_2013_01_31|OntoQA]] somewhere?

[10:33] Leo Obrst: @Till: I'm not sure. We'll query Samir Tartir and the other [[ConferenceCall_2013_01_31|OntoQA]] folks.

[9:52] anonymous morphed into Ken Baclawski

[9:54] Peter P. Yim: == Terry Longstreth / Todd Schneider presenting ... see: the [2-Track-B] slides

[10:04] Peter P. Yim: == Mike Bennett / Matthew West presenting ... see: the [3-Track-C] slides

[10:08] GaryBergCross: I'm surprised that people haven't mention the difference between evaluating

light weight ontologies vs. 'heavy" ones with lots of axioms.

[10:09] Maria Poveda: @Gary could it be the first one a subset of the second option?

[10:09] Maria Poveda: I mean in a heavyweight ontology you can make the mistake you do in a taxonomy

for example, and many other

[10:13] Simon Spero: @Maria: @What is the mistake of a taxonomy

[10:14] Simon Spero: @Maria: Apart from assuming that a taxonomy is necessarily an ontology

[10:14] Maria Poveda: @simon for example to set a subclassOf axiom wrong or to include a class in two

levels of the hierarchy

[10:14] Maria Poveda: @Simon, some authors also consider cycles as errors, they are not always, but

in some cases they are

[10:16] Simon Spero: Maria: Cycles are errors in controlled vocabularies other than synonym rings

[10:12] Simon Spero: @Gary How do you measure the weight of an ontology

[10:12] GaryBergCross: @maria It might be more useful to speculate that light weight ones are

transparent to the end user who have an understanding of those requirements rather than formal

requirements that a Knowledge Engineer understands as part of development.

[10:14] GaryBergCross: @Simon see http://en.wikipedia.org/wiki/Lightweight_ontologies for a

discussion....

[10:15] Simon Spero: @Gary: That's what I thought, but I've heard the, er, term misused

[10:17] GaryBergCross: @Simon Yes, I am more comfortable with the idea of a light ontology than a

heavy one, which is why I use quotes.

[10:27] Leo Obrst: @Gary et al: granularity is a factor distinguishing lightweight from heavyweight

ontologies, and is dependent on the intended application. If one does not need to distinguish

between 2 subclasses, e.g., one only needs to represent their parent class. This might be good for a

simple search and indexing application.

[10:33] GaryBergCross: @Leo Granularity might not be the right concept, but I think that I know what

you mean. The reason I don't think this as granular is that one might have 20 sub-types or parts or

scale levels in a light model but only get to 3-4 in a more formal one that is better modeled.

Concepts get detailed in more formal ontologies.

[10:39] Leo Obrst: @Gary: yes, you always have that distinction: a very large ontology could be very

simple, but cover a lot of simple classes, whereas a very precise ontology may just cover 3-4 of

those classes, so a kind of zooming in. It really is granularity and precision. Think of a map and

the actual region it maps.

[10:36] Maria Poveda: @Leo @Gary, could it be "expressivity"? I'n not sure about the term either...

[10:39] Maria Poveda: something like this

http://www-ksl.stanford.edu/people/dlm/papers/ontologies-come-of-age-mit-press-(with-citation)_files/image005.gif ?

[10:41] Leo Obrst: @Maria: Or this: ;)

http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2006_01_12#nidIT2. See slide 9, the Ontology Spectrum.

[10:46] Maria Poveda: @Leo thanks

[10:16] Todd Schneider: Leo, Steve. Terry, Has anybody discussed the notions of 'intrinsic' and

extrinsic' criteria w.r.t. particular lifecycle phases?

[10:18] Todd Schneider: It would seem that intrinsic criteria/evaluations are more relevant during

early lifecycle phases, while extrinsic are more relevant to later lifecycle phases.

[10:20] GaryBergCross: @Todd I can see it working the other way. One starts by having a lighter wt

ontology that used terms from the domain and is very extrinsic and only later is formalized. You

wind up with something in between.

[10:20] Michael Denny: @ToddSchneider Perhaps just the opposite. One may begin with concerns about

the domain semantics (extrinsic) and then worry about how best to implement those in a formal

ontology (intrinsic).

[10:21] GaryBergCross: @Michael +1 we agree.

[10:19] Steve Ray: @Todd: Agreed

[10:19] Terry Longstreth: @Todd Not necessarily more relevant, but certainly more available

[10:19] Fabian Neuhaus: @Todd. I think inconsistency might be relevant at any stage :-)

[10:20] Steve Ray: @Fabian: But you would have eliminated the inconsistency early in the lifecycle...

[10:21] Terry Longstreth: @Todd (rephrase) Intrinsics (of which I would include syntax checking) are

probably tested from the outset of a project.

[10:21] Todd Schneider: Fabian, Yes, but the manifestation of inconsistency will be different, hence

the associated evaluation criteria should be expected to be different.

[10:20] Fabian Neuhaus: @Steve: any change to the ontology might potentially lead to inconsistency.

That can happen at any stage of the life cycle.

[10:21] Steve Ray: @Fabian: Of course. I was half serious.

[10:22] Fabian Neuhaus: @Steve. Sorry. the humor got lost in the medium :-)

[10:23] Steve Ray: @Fabian: I'll try to use :) more often.

[10:21] Amanda Vizedom: @Todd: I don't think so; I've seen too many projects go off on the (wildly)

wrong track because they haven't defined their requirements (or shared them across sub-teams) up

front. I have been a pained neighbor-observer to one that when through three different start-end

year-long contracts like this, during each round of which the new contractor went off building

something with no potential to meet the need even if perfectly executed.

[10:22] Leo Obrst: @Fabian, Amanda: Yes, one might have a perfectly consistent ontology with good

structural metrics that is just nonsense, because it doesn't conform to the world. When we made this

a scale, we really weren't necessarily thinking of a 2D scale. It really was more of a rhetorical

device to talk about the space.

[10:26] Todd Schneider: Amanda, Michael Denny, There is the design phase, prior to any development,

where the issues you each raised would be addressed.

[10:28] Amanda Vizedom: @Todd: Yes, and it's important to make explicit that a good methodology

includes such a phase (design and/or requirements identification). Too many people never even think

of doing it for ontologies.

[10:23] Doug Foxvog: @Simon @Maria: cycles in a controlled vocabulary comprise what WordNet calls a

synset. In ontological terms, they are equivalence sets. One can have narrow contexts in which the

only instances of the more general concept are also instances of the narrower one. This can be

expressed in a domain ontology by creating a subclass cycle -- which defines all classes in the

cycle to be equivalent.

[10:24] Maria Poveda: http://oa.upm.es/6456/1/Evaluation_of_Taxonomic_K.pdf here there are some

example of what have been identify as errors in taxonomies

[10:24] GaryBergCross: Q. Do these evaluation approaches and concepts apply equally to Ontology

Design Patterns or are there additional factors to consider for ODPs?

[10:25] Amanda Vizedom: @Gary: Yes. ;-)

[10:25] Maria Poveda: yes, as in load of situations, if the goal of the developer is to do that it is

correct, the problem is when people (most of the time not ontologists) end up with that models by mistake

[10:26] Maria Poveda: so IMHO I would not look for things that are always and error because it is

going to be almost impossible because some one might want to do that

[10:26] Maria Poveda: but find situation that might be an error, identify them and decide whether

they should be corrected or not

[10:27] Peter P. Yim: == Michael Denny / Peter P. Yim presenting ... see: the [4-Track-D] slides

[10:36] Amanda Vizedom: @MichaelDenny: I agree and think your point is important; it seems as though

many (most?) evaluation factors will be relevant at many (most?) points in the lifecycle, but

perhaps call for different treatment at different points.

[10:40] Amanda Vizedom: @MichaelDenny: (ref. MichaelDenny's remarks during the presentation, that he

may not consider "visualization" a factor in the software that addresses ontology quality) Visual

building, or visualization, might be important for developer understanding OR for SME validation.

The survey could ask about this kind of capability at different levels of specificity, though (i.e.,

render ontology in format intelligible to non-ontologists and facilitate input or review from same

vs. "ontology visualization" and/or "visual ontology construction."

[10:40] Terry Longstreth: @Amanda +1

[10:42] Mike Bennett: @Amanda +2. Presentation to SMEs may take one of several forms, since non

technical people, being people, have one of several modalities they may be comfortable with: visual

Boxes and lines per Vision/whiteboard; tables or spreadsheets of terms and definitions; and

controlled natural language.

[10:43] Terry Longstreth: about extrinsic vs intrinsic (or the hybrid) is inspecting and judging a

visual representation of the CLIF based ontology an extrinsic or intrinsic test?

[10:56] Amanda Vizedom: @Terry: it depends what you are inspecting and judging it for.

[10:43] Peter P. Yim: == Open Discussion about the tracks' syntheses ...

[10:52] Steve Ray: @Fabian: It's hard to imagine good performance for an ontology-based system if the

ontology doesn't align with reality.

[10:52] Doug Foxvog: @Leo: the need is not always conformance to the "real world", but conformance to

the world of the domain. One could create an ontology for a role-playing game that uses ontology for

real-world physics, but adds fictional biological creatures, and fantastical powers.

[10:53] Doug Foxvog: @Steve: see my above statement.

[10:53] Steve Ray: @Doug: Yes, let me amend my use of the word "reality" to mean "the domain being modeled"

[10:56] Leo Obrst: @Doug: yes, this gets into the "reality" of the domain, and so might be better

considered as verisimilitude to the domain, if the domain is e.g., fictional.

[10:55] Amanda Vizedom: @Steve: Some "ontologies" do fairly well in their initial the context for

which they are originally created, because of implicit assumption shared in that context, but are

impossible to reuse in part because of basic problems, including conformance to reality *when just

the explicitly captured ontology is considered*.

[10:57] Steve Ray: @Amanda: What you describe sounds to me like the ontology is simply incomplete, if

there are assumptions that are not reflected in the model. That, to me, is different from not being

aligned with the domain being modeled.

[10:57] Mike Bennett: All these considerations about the reality of the domain, have two possible

implications: quality criteria for the thing you are developing for; and considerations when others

are developing something and want to consider whether or not to reuse that ontology.

[10:58] Ali Hashemi: @Steve - this comes back to the point of granularity and precision, right?

[10:58] Amanda Vizedom: @Terry: *assuming*, that is, that the visualization technique preserves the

structure and content of the ontology.

[10:58] Ali Hashemi: a lightweight ontology will almost certainly leave many things left unsaid

[10:59] Steve Ray: @Ali: Yes, I agree. All of this falls within the "correctness" of an ontology,

IMHO.

[10:59] Ali Hashemi: From the verbal discussion that ensued, it's unclear to me how the "conformance

to the reality of the target domain" is reflected in our tracks.

[10:59] Doug Foxvog: @Ali: *Every* ontology will certainly leave many things unsaid.

[10:59] Ali Hashemi: @Doug, agreed. So I'm not sure it's useful to simply state that an ontology is

complete if they didn't "completely" model their domain

[10:59] Steve Ray: @Ali: See my comment at [10:52]

[11:00] Ali Hashemi: incomplete*

[11:00] Mike Bennett: Apologies, I have to drop off now.

[11:01] Terry Longstreth: An important reason to continue to talk about implicit vs. explicit is to

minimize the opportunity for misinterpreting results. If the ontology picture (from my [10:14]) is

judged by the SME to be logically correct and aesthetically pleasing, it still needs to be validated

against the CLIF representation for logical equivalence, and the CLIF version has to be

intrinsically evaluated, to put bounds on the range of valid inferences that can be drawn from it.

[11:02] Amanda Vizedom: @Steve: original use might not have included machine reasoning; artifact may

therefor have inconsistencies in the explicit content that aren't caught (the original users may

interpret not according to the formal semantics, but rather according to some conventional treatment

local to them. They often don't realize this and put their artifact out there as a reusable ontology.

As soon as the formal semantics are applied, e.g., by machine validation, inconsistencies are apparent.

[11:04] Steve Ray: @Amanda: In that case, the incompleteness is in the evaluation.

[11:05] Amanda Vizedom: @Steve, in the initial evaluation, yes. When a new party considers this

artifact for reuse, *their* evaluation should catch it.

[11:02] Doug Foxvog: Note that in a *knowledge base*, that uses an ontology, data need not be limited

to identification and properties of individuals. Additional properties of classes and rules for the

context can also be defined.

[11:04] Terry Longstreth: @SteveRay : you might want to say correctly constructed, or perhaps well-formed.

[11:05] Ali Hashemi: @Terry, I like the notion of a well-formed ontology

[11:11] Terry Longstreth: Why concern for well-formedness? It's necessary to insure that the semantic

properties/extrinsics are correctly represented

[11:06] Doug Foxvog: @Amanda: It's true that inconsistencies can be made apparent when formal

semantics are applied. But these can often be "automatically" determined when missing statements

(e.g., disjointness) are specified. The Cell Line Ontology until Dec. 2011 had a number of common

subclasses of PlantCell and AnimalCell (and likewise of other disjoint class pairs). Upon specifying

disjointness in a reasoning system, these all popped out.

[11:08] Doug Foxvog: A system can count the use of disjointness and similar assertions as indicators

and suggest that low use indicates incompleteness of an ontology.

[10:55] Peter P. Yim: == Fabian Neuhaus / Amanda Vizedom presenting ... see: the [5-Communique] slides

[11:03] Steve Ray: @Fabian: (Regarding the communique). On slide 6, you don't seem to include "Is the

ontology well constructed?" which is what all the intrinsic evaluation is about.

[11:04] Ali Hashemi: @Fabian, will there also be a finer-grained discussion of how the various

metrics and evaluation approaches map to the various types of ontologies?

[11:07] Peter P. Yim: == Open Discussion about the Communique Approach ...

[11:08] Matthew West: One of the things that I think that would be very useful in structuring the

Communique is the use of a net of problems/net of solutions approach. It can help to gather together

and sift through the mass of detail we have.

[11:09] Amanda Vizedom: @Matthew: thanks for that suggestion; we will look into that.

[11:15] Matthew West: You can find a very simple net of problems here on Page 12

http://www.matthew-west.org.uk/documents/princ03.pdf A simple net of solutions is on P40.

[11:09] Todd Schneider: Amanda, Fabian, Were does the notion of lifecycle come in?

[11:19] Todd Schneider: To provide value this summit should provide some guidance as to when a

particular evaluation/validation, or collection there of, should be applied w.r.t. the system

lifecycle.

[11:10] Ali Hashemi: @All - will we be referring back to the Ontology Usage that was developed in the

previous Summits? This would help select subsets of the very broad ranges of tools, approaches and

metrics developed in this summit.

[11:11] Ali Hashemi: cf -

http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011_ApplicationFramework_Synthesis

[11:12] Fabian Neuhaus: It seems the call was dropped

[11:13] Doug Foxvog: The conference hung up in the middle of Amanda's talking

[11:13] Steve Ray: Did everyone just lose a connection? I suppose so.

[11:13] Simon Spero: Please hold. operators are standing by.

[11:13] Leo Obrst: Yikes! Peter must be dropped.

[11:13] Amanda Vizedom: Hopefully Peter will return soon...

[11:13] Steve Ray: Calming music.

[11:13] Doug Foxvog: We're back

[11:13] Simon Spero: Everyone stop talking about Peter

[11:14] anonymous morphed into Lamar Henderson

[11:14] Amanda Vizedom: My point is: I think the lifecycle will appear most in the context of best practices.

[11:16] Doug Foxvog: Should lifecycle be orthogonal to ontology validation/verification feature? If

not, could we specify which components are needed at which points in the lifecycle?

[11:20] Amanda Vizedom: @DougFoxvog, we might, but we would not want to try to enumerate this in the

Communique. This is similar to the question of what evaluation factors are relevant to what sorts of uses.

[11:20] Michael Denny: @DougFoxvog This is what the survey attempts to do by sorting some 100

software capabilities adressing ontology quality/fitness into seven phases of the lifecycle.

[11:22] Amanda Vizedom: Both are areas of high value. I hope we make progress on them during the

Summit. I don't think we can or should try to enumerate all specifics of such factors in the

Communique. The findings and conclusions, however, should be reflected in the best practices and

future content.

[11:22] Ali Hashemi: @Amanda, might they be enumerated on the synthesis pages?

[11:23] Amanda Vizedom: @Ali, yes, I hope so. They might also turn into additional summit outputs, in

one form or another!

[11:14] Peter P. Yim: == Peter P. Yim / Mike Dean presenting ... see: the [6-Hackathon-Clinics ] slides

[11:19] Anatoly Levenchuk: We issued today version 1.2 of .15926 Editor (this is like Protege for ISO

15926 ontology, while RDF/OWL is also supported): http://techinvestlab.ru/dot15926Editor -- it will

be used during hackathon/clinics for multiple ontology evaluation activities within our

Russian-speaking community (http://dot15926.livejournal.com/40280.html -- hackathon/clinics

announced in Russian).

[11:21] GaryBergCross: Evaluating an ontology developed by building out from an ODP might be an

interesting exercise.

[11:23] Doug Foxvog: I note that sending "gold standard" ontologies through evaluation tools may very

well determine problems with the ontologies labeled "gold standard". Such use will not only be

useful for evaluating the tools, but also the "gold standards".

[11:25] Terry Longstreth: @Peter: I'd be careful with the term SME. It conjures images of greybeards,

but in an IT environment, the people most interested in the correctness of function are usually

called operators or users.

[11:25] David Leal: The CEN SERES workshop will produce an ontology for materials data before May. We

have participants also involved with the US Material Genome Initiative, so this ontology may have a

broad role. The ontology may be presented as an extension of ISO 15926. It would be very good if the

ontology validation tools could be used on this ontology.

[11:28] David Leal: Materials ontologies are a bit of a challenge for industry understandability,

because the word "material" is used to mean both batch of stuff and a type of stuff - sometimes in

the same sentence.

[11:29] GaryBergCross: Earth Cube is interested in various kinds of "material entities."

[11:31] Doug Foxvog: @DavidLeal: Materials ontologies need to realize that some properties of

materials are properties of the "stuff" independent of the state of affairs (temperature, pressure,

...), others are "intrinsic" as they do not depend upon the amount of material, but may depend on

temp/pressure/other environment, while others are extrinsic, depending on the physical object

[11:31] David Leal: and the depend upon the history of the environment - materials have memory!

[11:32] Doug Foxvog: @David: the material memory is in its microscopic structure. Since that is hard

to specify, specifying the history of events helps.

[11:26] GaryBergCross: It might be interesting to take a fairly loose, prototype or light ontology

and see if the evaluation gives us a way of understanding how to forge it into something good. Sort

of agile engineering approach.

[11:27] GaryBergCross: @ Amanda..perhaps we can work up the Hydro model to a form that would be

submitted for a test.

[11:29] Amanda Vizedom: @Gary: Seems like a good candidate.

[11:30] Amanda Vizedom: @Gary ... we'd need to get unstuck on the use case specification.

[11:30] GaryBergCross: @Amanda Let us see if we can get USGS interested in moving forward from their data.

[11:28] Leo Obrst: Wow, we certainly have a lot here!

[11:30] Peter P. Yim: == Q&A and Open Discussion ...

[11:31] Maria Poveda: bye :-)

[11:33] Peter P. Yim: join us again, same time next week, for Ontology Summit 2013 session-07: "Extrinsic

Aspects of Ontology Evaluation - II" - Co-chairs: Todd Schneider & Terry Longstreth -

http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_02_28

[11:33] Joanne Luciano: thanks!

[11:33] Peter P. Yim: -- session ended: 11:31 am PST --

-- end of in-session chat-transcript --

  • Further Question & Remarks - please post them to the [ ontology-summit ] listserv
    • all subscribers to the previous summit discussion, and all who responded to today's call will automatically be subscribed to the [ ontology-summit ] listserv
    • if you are already subscribed, post to <ontology-summit [at] ontolog.cim3.net>
    • (if you are not yet subscribed) you may subscribe yourself to the [ ontology-summit ] listserv, by sending a blank email to <ontology-summit-join [at] ontolog.cim3.net> from your subscribing email address, and then follow the instructions you receive back from the mailing list system.
    • (in case you aren't already a member) you may also want to join the ONTOLOG community and be subscribed to the [ ontolog-forum ] listserv, when general ontology-related topics (not specific to this year's Summit theme) are discussed. Please refer to Ontolog membership details at: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
      • kindly email <peter.yim@cim3.com> if you have any question.

Additional Resources


For the record ...

How To Join (while the session is in progress)

Conference Call Details

  • Date: Thursday, 21-Feb-2013
  • Start Time: 9:30am PST / 12:30pm EST / 6:30pm CET / 17:30 GMT/UTC
  • Expected Call Duration: ~2.0 hours
  • Dial-in:
    • Phone (US): +1 (206) 402-0100 ... (long distance cost may apply)
      • ... [ backup nbr: (415) 671-4335 ]
      • when prompted enter Conference ID: 141184#
    • Skype: joinconference (i.e. make a skype call to the contact with skypeID="joinconference") ... (generally free-of-charge, when connecting from your computer)
      • when prompted enter Conference ID: 141184#
      • Unfamiliar with how to do this on Skype? ...
        • Add the contact "joinconference" to your skype contact list first. To participate in the teleconference, make a skype call to "joinconference", then open the dial pad (see platform-specific instructions below) and enter the Conference ID: 141184# when prompted.
      • Can't find Skype Dial pad? ...
        • for Windows Skype users: Can't find Skype Dial pad? ... 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 on 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. ... (ref.)
  • Shared-screen support (VNC session), if applicable, will be started 5 minutes before the call at: http://vnc2.cim3.net:5800/
    • view-only password: "ontolog"
    • if you plan to be logging into this shared-screen option (which the speaker may be navigating), and you are not familiar with the process, please try to call in 5 minutes before the start of the session so that we can work out the connection logistics. Help on this will generally not be available once the presentation starts.
    • people behind corporate firewalls may have difficulty accessing this. If that is the case, please download the slides above (where applicable) and running them locally. The speaker(s) will prompt you to advance the slides during the talk.
  • In-session chat-room url: http://webconf.soaphub.org/conf/room/summit_20130221
    • instructions: once you got access to the page, click on the "settings" button, and identify yourself (by modifying the Name field from "anonymous" to your real name, like "JaneDoe").
    • You can indicate that you want to ask a question verbally by clicking on the "hand" button, and wait for the moderator to call on you; or, type and send your question into the chat window at the bottom of the screen.
    • thanks to the soaphub.org folks, one can now use a jabber/xmpp client (e.g. gtalk) to join this chatroom. Just add the room as a buddy - (in our case here) summit_20130221@soaphub.org ... Handy for mobile devices!
  • Discussions and Q & A:
    • Nominally, when a presentation is in progress, the moderator will mute everyone, except for the speaker.
    • To un-mute, press "*7" ... To mute, press "*6" (please mute your phone, especially if you are in a noisy surrounding, or if you are introducing noise, echoes, etc. into the conference line.)
    • we will usually save all questions and discussions till after all presentations are through. You are encouraged to jot down questions onto the chat-area in the mean time (that way, they get documented; and you might even get some answers in the interim, through the chat.)
    • During the Q&A / discussion segment (when everyone is muted), If you want to speak or have questions or remarks to make, please raise your hand (virtually) by clicking on the "hand button" (lower right) on the chat session page. You may speak when acknowledged by the session moderator (again, press "*7" on your phone to un-mute). Test your voice and introduce yourself first before proceeding with your remarks, please. (Please remember to click on the "hand button" again (to lower your hand) and press "*6" on your phone to mute yourself after you are done speaking.)
  • RSVP to peter.yim@cim3.com with your affiliation appreciated, ... or simply just by adding yourself to the "Expected Attendee" list below (if you are a member of the community already.)
  • Please note that this session may be recorded, and if so, the audio archive is expected to be made available as open content, along with the proceedings of the call to our community membership and the public at-large under our prevailing open IPR policy.

Attendees

  • Expecting:
    • ...
    • please add yourself to the list (above) if you are a member of the Ontolog or Ontology Summit community, or, rsvp to <peter.yim@cim3.com> with your affiliation.