ppy/chat-transcript_edited_20110331b.txt Chat transcript from room: ontolog_20110331 2011-03-31 GMT-08:00 --------------- PeterYim: Welcome to the ... OntologySummit2011: Panel Session-9 - All Hands: "Synthesis and Reports" - Thu 2011_03_31 Summit Theme: OntologySummit2011: Making the Case for Ontology Session Title: All Hands: Synthesis and Reports Session Co-chairs: Dr. SteveRay & Dr. NicolaGuarino Panelists: Summit Co-chairs, Track Co-champions and Communique Co-Lead Editors (* = presenting today) * OntologySummit2011 General Co-chairs ... SteveRay* & NicolaGuarino * Track-1: Ontology Application Framework ... MichaelGruninger* & MichaelUschold & NicolaGuarino * Track-2: Applications and Case Studies ... MillsDavis & MikeBennett* * Track-3: Value Metrics, Value Models and the Value Proposition ... ToddSchneider* & RexBrooks * Track-4: Strategies for "making the case" ... MatthewWest* & PeterYim * Track-5: Grand Challenges ... RamSriram* & ErnieLucier & AldenDima * Communique Co-Lead Editors ... MichaelUschold*, JohnSowa, MillsDavis & JohnBateman Please refer to details on the session page at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2011_03_31 . anonymous morphed into BrianHaugh anonymous morphed into MatthewWest anonymous morphed into ChristopherSpottiswoode RexBrooks: I'm stil getting "incorrect conference id" I use Vonage and this is the first time this has happened. ... I even rebooted the device. But no change. MikeBennett: @Rex w some phones, have to hold down keys for 1 sec each. JohnBateman: I got the incorrect conference id a couple of times, then it worked RexBrooks: Finally got through! PeterYim: @RexBrooks, MikeBennett, JohnBateman, MichaelGruninger et al. - the conference bridge is a bit tricky ... you almost has to space out the keystrokes evenly (especially with the # sign at the end) to have it work well - see: http://ontolog.cim3.net/cgi-bin/wiki.pl?VirtualSpeakerSessionTips#nid2OR0 MikeBennett: OK, I'm going to have to redial in PeterYim: MichaelGruinger is on line now MichaelGruninger: I'm online, but I cannot unmute PeterYim: @MichaelGruinger - let me unmute you, please try your voice when you have a chance anonymous morphed into DavidLeal SteveRay: Standard cheat sheet: *2 to mute, *3 to unmute PeterYim: MichaelGruinger: additionally, will also incorporate the ontology classifications that OMV has provided NicolaGuarino: Anyway, I'll state my question on the chat: my problem is related to the very name "ontology application framework" PeterYim: discussion below is in reference to a possible change of the Track-1 label to disambiguate the possibility that the audience may misinterpret us (as in the case of StefanoBertolo, who thought we were trying to find out if something like that would be a good candidate for research funding) ToddSchneider: @Nicola, the Ontology Spplication Framework (OAF) may be at too low a level for stakeholders/decision makers. MichaelGruninger: Would "Taxonomy of Ontology Applications" be better JohnBateman: but taxonomy is bad! AmandaVizedom: I think "taxonomy" makes certain commitments about the structure/relationship of the feature, and should be avoided NicolaGuarino: I agree it is not a taxonomy SimonSpero: "What Can Ontologies Do For You?" JohnSowa: How about a very short title: Applications. ToddSchneider: John, simple is hard. JohnSowa: The word 'ontology' can be deleted, because that's the overall title. JohnSowa: The word 'usage' can be deleted, because every application is used. JohnSowa: When you delete all the problematical words, the only one left is 'application'. TimWilson: Is that a directive from the Department of Redundancy Department? RexBrooks: I think that Ontology USE Framework captures it more succinctly. MikeBennett: @Rex +1 TimWilson: @Rex +2 ToddSchneider: Michael, the 'application' level may still be too low level, at least for an 'elevator' pitch. NicolaGuarino: Ontology applications comparison framework? MichaelRiben: Is the group familiar with the caBIG's effort to develop a similar ontology app framework which they call the Ontology "representation": https://wiki.nci.nih.gov/display/VCDE/Ontology+Representation+WG SimonSpero: Applications, or Problems? AmandaVizedom: Using the language of the Wiki page: Ontology Applications Description Framework MichaelUschold: ON ambiguity of the term "ontology application framework" I agree that application is ambiguous here. How about ontology deployment framework? NicolaGuarino: @MichaelUschold: +1 ChristopherSpottiswoode: Ontology Application Typology? SimonSpero: "What kind of problems can ontologies solve?" TimWilson: There appears to be little consideration for the use of ontology to augment search, particularly in the enterprise. MikeBennett: Application has 2 meanings MikeBennett: Crux of the problem is "Application" as a kind of program v. Application as "how something is applied" SimonSpero: MikeBennett+ RexBrooks: Perhaps Ontology Usage Framework? MikeBennett: Better RexBrooks: We're capturing the kinds of uses to which ontology can be put. MatthewWest: Most of the ambiguity comes from missing prepositions. Framework for the Application of Ontologies would (for example) get over that. RexBrooks: @Matthew: yes. JohnSowa: Long titles require more explanation than they clarify. AmandaVizedom: "Framework for Understanding Ontology Applications?" MichaelRiben: is the purpose for "discovery" in order to know if the person can apply it to their use case? ToddSchneider: +1 for 'Guide' NicolaGuarino: Let's delegate to the champions to choose the final title on the basis of the input received PeterYim: I agree with Nicola ... let's defer to the Track-1 champions AmandaVizedom: @NicolaGuarino: +1 ... and it's then most useful for the rest of us to comment/throw out idea in this chat for them to consider later. SimonSpero: (past half hour. Now afraid) SimonSpero: [Changed Mikes - Now on track 2 - Use Cases] MichaelRiben: I think a good healthcare/public health example would be to look at Parsa Mirhaji's Sapphire system for the CDC that performs surveillance in the Houston area...see http://knowmed.com/Solutions.html MichaelGruninger: Should we think of the Ontology Application Framework as really being "Ontology Use Case Metadata"? The idea is that all of the dimensions provide the terminology for describing a use case from a technical and (eventually) business perspective. SteveRay: I think using the word "metadata" is just going to confuse things. MichaelUschold: In discussing benefits of an ontology application, it is critically important to link back to how the ontology makes that possible. Saying 'customer retention' gets people's attention, but you need to explain how the ontology helped to do this. AmandaVizedom: @MikeBennett: +1 for adding the dimension of What's Modeled. Not only do people differ on this, but a lack of understanding of that difference has caused noticeable confusion and disagreement on list during this summit. It's surely a dimension of difference that affects best practices and measures of quality! MikeBennett: @Amanda +1 SimonSpero: MikeBennett: were the metrics that you hadn't heard of internal to the use case presenter? MikeBennett: @Simon yes, it was some CRM acronym MikeBennett: So, what does AHT stand for? (in CRM / call center context)? MikeBennett: @Simon - AHT was my unexplained metric from CRM case study, as now noted in Track 3. SimonSpero: @MikeBennett: Either Average Hold Time or Average Handling Time (http://www.kellen.net/crm_mf.pdf) MikeBennett: @Simon, thanks. Not my guess, so worth defining (terminology!). SteveRay: @MikeBennett: I want to ensure we are going to have pointers to as many actual cases as possible. Is that your intention? MikeBennett: @MichaelUschold @SteveRay yes, I need to take the assertions in these slides and link to the relevant case study. On the Wiki. TimWilson: I was hoping to stay on the call for Matthew and Peter's presentation, however I must get back to work. I will be presenting to my superiors next week for the creation of a knowledge initiative using ontology. TimWilson: This summit has been invaluable to me in that effort. RexBrooks: It just dawned on me that one area where we haven't seen specific Use-Cases is internal corporate governance and management both IT Management and Business Intelligence Management. MichaelRiben: need to go..sorry PeterYim: @MichaelRiben - thanks for joining, and for the contribution SimonSpero: @MikeBennett: Do you have the data coded in a way that it could be fed into Multi Dimensional Scaling for graphing? MikeBennett: @Simon re feeding graphing:no. not many numeric measures were asserted. RexBrooks: JohnSowaa noted the connection between people using terms in human interoperability and the improvements that come about due to that. NicolaGuarino: Excellent question, John a crucial one! SteveRay: @John: Fascinating point about getting humans to be interoperable, a la common terminology. I am definitely running into that in the world of disaster management interoperability, where "interoperability" is commonly interpreted as "interchangeability of emergency personnel" ToddSchneider: Steve, look at the NCOIC SCOPE model to get a better understanding of 'interoperability'. ToddSchneider: One perspective of interoperability is crossing boundaries (that didn't need crossing in the past or haven't been considered explicitly in the past) SimonSpero: [summary of discussion: labels/words are for people ; MikeBennett points out that business users are concerned about definitions] AmandaVizedom: @JohnSowa: But even in a case that appears to be as you say, nailing down and tracking the relationship between the different concepts *independently of the local terminology* or *across local terminologies* is typically critical to the getting the benefit. That is, the advantage is in being able to keep *both* local-human-interpretable presentation/usability *and* machine usability. MichaelUschold: @JohnSowa - there is a lot of ambiguity in natural language definitions, because we all use terms in different ways. There is also a lot of ambiguity in formal logic definitions because they miss so much of the intended meaning, which is more easily expressed in natural language. NicolaGuarino: @JohnSowa (I repeat it because there was a typo): there is a (huge) benefit related to the terminology used, ONE related to the way the formalization of the intended meanind is done (independently from computational aspects), and finally one related to the computational use of the terminology (typically for taxonomic reasoning) NicolaGuarino: @MichaelGruninger & MichaelUschold: maybe JohnSowa's point on distinguishing between different uses of ontologies (terminological clarification vs. formal clarification vs. automated reasoning) can be made explicit in the "application framework"... MichaelGruninger: @Nicola: Yes MichaelUschold: Re: @Nicola's comment about incorporating @JohnSowa's distinction. Agree, this should be incorporated. To some extent it is there already. SimonSpero: [My diss. topic is on interface between controlled vocabularies and ontologies. It's been misunderstood by people on both sides since before Aristotle was a gleam in his father's eye] BartGajderowicz: Regarding controlled vocabularies, there is a survey out about this. Perhaps beneficial to check back on the results: http://semanticweb.com/finding-out-whether-controlled-vocabularies-matter_b18740 MichaelUschold: On Value metrics: if a value model is a subset or subtype of another one, maybe a good idea is to show this in a venn diagram or tree structure. SimonSpero: [Now moving to track 3 - Value Models, Value Metrics and Value Proposition] PeterYim: ALL: please identify clearly who you are addressing with "@" before your message (like @MichaelGruninger or @MichaelUschold, @JohnSowa or @JohnBateman, so that it will not be ambiguous) - I need that help because in the post-processing of the chat transcript, I will reshuffle the dialog so that Questions and Answers, and related discussion are placed adjacent to one another RexBrooks: @ToddSchnieder, I believe you were referring to the granularity matching between use case and metrics. MichaelUschold: Some of @ToddSchneider's material on value models and metrics could be integrated into the ontology application framework, rather than standing alone. RexBrooks: Yes! SimonSpero: [track change - track 4 - Strategies for "Making the Case"] SimonSpero: [weren't FFC's actors often drunk, on LSD, or both?] AlanRector: The speaker keeps fading out RexBrooks: Yes, we need Matthew to speak a little louder. RexBrooks: The granularity issue applies to the strategy, too. If we narrow our focus to specific use-cases or the most generic use cases, we stand a better chance of scoring well. RexBrooks: The notion of which values count most is a good consideration. BartGajderowicz: Is there any benefit to talking about sub-ontologies, micro-theories (Cyc term) or axioms in repositories (COLORE) vs. full ontologies? Full ontologies may be overkill for most applications? MikeBennett: @Bart one or 2 of the case studies showed benefit in reuse of existing, common ontology concepts RexBrooks: @Bart: Not sure I would use those kinds of arguments to any audience other than other ontologists or folks with an interest in ontology. For the end users I would focus on solving their problems and only bring up HOW if asked. SimonSpero: It turns out that Data Quality has to have a case made for it too, alas. [ http://portal.acm.org/citation.cfm?id=1534235 ] JohnBateman: @BartGajderowisz: definitely, but only when the audience has got out of the elevator, into the meeting room and then onto the details... so a question of granularity in what level of pitch when I'd guess. Part of the methodology for engineering things properly (so it is a selling point that there *is* a methodology). NicolaGuarino: @Matthew: speaking of the role of ontology on data quality, it occurs to me that in many cases just "ontology awareness" is enough to produce good data (e.g., choosing good, non-ambiguous names for the entities we talk of), without necessarily using ontologies as artefacts. This means that a certain ontology education is necessary for building the controlled vocabularies JohnSowa mentions... MatthewWest: @Nicola: Yes. You have relatively little chance of success without an ontological approach, even if that is not what it is called. AmandaVizedom: @Bart - this touches another angle, of great interest to me, from which the framework can be used: the need to better understand ontology best practices and principles of good ontology design. There is very little work done that reflects or addresses the way good practices depend, in many ways, on the application context. Seeing how this framework might help us understand those dependencies is a direction of great promise, I think! BartGajderowicz: @AmandaVizedom: Yes. I've notices some work on analyzing adoption of Semantic Web technologies, now that there is enough information. There's a few reports I ran across. I need to dig them up and can forward. BartGajderowicz: @MikeBennett: Thanks, I saw the slides on reuse. I was referring more to the size of the ontologies. For example, FOAF and GoodRelations are relatively small but very specific in their scope, which (assuming) helped with their adoption. MikeBennett: @Bart there is definitely a case to be made for a set of common, granular ontologies that people can refer to (time, geog, accounting, legal etc.). BartGajderowicz: @RexBrooks: same answer. Smaller ontologies are specific to a particular problem, so more relatable to the end user. RexBrooks: @Bart: agreed. MikeBennett: @Bart the audience to whom to make that case is the Linked Data folks I think BartGajderowicz: @MikeBennett: Agreed. But I view sub-ontologies as being used in analyzing data, not just linking it. When I look at the Dataset Publishing Language (DSPL) project, "Slices" look like different combinations of relationships between "Concepts". These "Slices" resemble sub-ontologies to me. http://code.google.com/apis/publicdata/docs/tutorial.html MikeBennett: @Bart interesting. thanks for the pointer. One thing we have done in the EDM Council work is identify common terms and commit to trying to find the most appropriate standard for each (e.g. ISO 3166 for countries). BartGajderowicz: @MikeBennett: Thanks. I will take a look at this work. MikeBennett: MatthewWest's Slide 10 is a better list of the kinds of people we should make the case to, than what I had in the Case Studies similar slide. MatthewWest: @MikeBennett: I notice there is considerable crosstalk between the tracks. Ours in particular, since it is so general. You might want to pick up MDM as a case study too. PeterYim: [track change - track 5 - Grand Challenges] MichaelUschold: I agree with @JohnSowa's voice comment that the grand challenge problems are more general AI problems than ontology problems per se. RexBrooks: @JohnSowa: Agreed. A Grand Challenge would be to "Make the Cases for Ontology" highlighting the "Cases". MichaelUschold: On the Grand Challenge track. While it is useful and interesting overall, I do not see a clear connection to how to make a case for ontology. What has been learned in this track that can be distilled into a tip, hint or guideline that helps make a case for ontology? SteveRay: @MikeUschold: I think this track simply provides a pointer to future opportunity areas where one could apply the methods we are articulating. PeterYim: @JohnSowa @MichaelUschold - as Matthew pointed out (slide#3) there is effective, NO CASE for Ontology (as was pointed out analogously during the Track-4 session, that there is no case for calculus) ... but the Grand Challenges will be able to focus our energy JohnBateman: @MichaelUschold: I guess this is along the lines: these applications are fantastic, to do them ontologies are a must, therefore this is the case for ontology... But I wonder if we can find applications which really are completely irresistable. NicolaGuarino: John: +1 MatthewWest: @JohnBateman: The things you are mentioning are research goals we need to identify. NicolaGuarino: @RamSriram: Speaking of great challenges, I would like to point to http://www.futurict.ethz.ch/FuturICT. FuturICT is a proposal for a 10-years European "Flagship" project, which scored first in a recent call, aiming at world-wide simulation, data interpretation, crisis management, etc. Great opportunities for ontologies, especially for ontology of large scale socio-technical systems MatthewWest: The thing about grand challenges is that they have some objective, you know when you have succeeded. I think an important prequel to grand challenges are research goals. What are the current gaps? A useful Grand Challenge would then depend on the achievement of some of those research goals. JohnBateman: I doubt there is going to be a grand challenge for *ontology* that is of interest to outsiders; the only challenges will *have* to be specific problems where ontologies can be argued to provide an important part of the work: as in FuturICT. BartGajderowicz: @JohnBateman: Yes. The granularity is much more relatable to end-users. You basically speak their language. The pitch woudl be something like " sub-ontology A is about geo, sub-ontology B is about time, etc. One of these may fit you domain. AldenDima: @JohnBateman I would also agree with you. We are asking potential "customers" for grand challenges and they don't per se care about ontologies. So I think we are running the risk of asking for challenges and then saying "Sorry not what we were looking for". But perhaps we should look for multidisciplinary challenges in which we can play a significant role and then partner with the other disciplines. Ramdsriram: @John: The purpose of the track is to identify a few grand challenges computational problems and identify opportunities for ontologies. For example, can we achieve the "kind of things" that Eliot, Ramesh, and Nabil talked about without using any ontologies (or related technology). SimonSpero: Is there an elevator in the NIST building? AldenDima: @Simon - we do have elevators in our buildings. I believe that the room for the Summit is on the ground floor. SimonSpero: AldenDima: I think that testing an elevator speech in an NIST elevator is pretty good metric AldenDima: @Simon - we can start with the elevator in the tall admin building and work our way to the shorter ones... ToddSchneider: All, I have to go. RexBrooks: Bye Todd. You gave a fine presentation, Todd. PeterYim: [track change - Communique] MichaelUschold: I'm getting on my cell, two seconds, hang on MikeBennett: @MichaelUschold (re slide 4) but does "the Ontology community" have one understanding of what it means and what it's used for? RexBrooks: @MichaelUschold: I think this is a terrific way to go and I look forward to the interim and final versions. MikeBennett: @Matthew indeed. Master data management (as distinct from metadata management - had me going!) is a feature of some of our case studies e.g. industry standard terms across an industry gets deployed in a number of ways within firms including master data (golden source), metadata repositories and other application areas. PeterYim: @MichaelUschold - just to make sure nothing falls through the crack, can you identify exactly who has promised you text by end-of-day Monday, please? NicolaGuarino: MichaelUschold's scheme is GREAT, I am just a little bit concerned about its level of ambition... MikeBennett: I promised text my end of Monday MatthewWest: I did Ramdsriram: @Peter: We should have a rough draft (hopefully). RexBrooks: I'm pretty sure Todd has promised Track 3 by end of Monday. If not I will speak with him. TerryLongstreth: Put the email address on the chat PeterYim: @Steve - ref. the dinner, the way you did it last year (a email to the list with a link someone can click on to provide you the response, offline) that would be great! PeterYim: great session ... very productive! JoelBender: thank you! NicolaGuarino: I agree, very productive and promising! PeterYim: -- session ended: 11:36am PDT -- ---------------------