ppy/chat-transcript_edited_20110203c.txt PeterYim: . Welcome to the OntologySummit2011: Panel Session-2 - Ontology Application Framework - I - Thu 2011_02_03 Summit Theme: OntologySummit2011: Making the Case for Ontology Session Title: Strawman for the Ontology Application Framework Session Co-chairs: Professor MichaelGruninger (U of Toronto) & Dr. MichaelUschold (Semantic Arts) Panelists: * Dr. MichaelUschold (Semantic Arts) - "A Framework for Understanding and Classifying Ontology Applications" * Professor MichaelGruninger (U of Toronto) - "Notes for an Ontology Application Framework" * Dr. LeoObrst (MITRE) - "Ontologies & Applications" * Professor AlanRector (U of Manchester) - "Problems arising in applications: Background Knowledge Representation, Data modelling & Ontologies" . Please refer to session details (dial-in, agenda, slides, etc.) at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2011_02_03 . anonymous morphed into PatBarkman anonymous morphed into ToddSchneider anonymous1 morphed into BruceBray anonymous2 morphed into AmandaVizedom anonymous morphed into RamSriram Susan Turnbull (GSA) morphed into SusanTurnbull anonymous morphed into AntoinetteArsic AlanRector: Please put details of paper on chat show - the line looses some detail - thanks PeterYim: @MichaelUschold - it's slide#2 (not #1) now PeterYim: slide#15 (not #14) YuLin: where is it now? I got lost PeterYim: For some reason, our speakers haven't been calling out their slide numbers very accurately, therefore (especially asynchronous participants) please bear with the situation, and try to sync up to the content as they are presented ArturoSanchez: @MichaelUschold: the main issue with all the approaches, as I understand them, is that semantic information that drive the mapping is not explicitly represented, and therefore needs to be mediated by humans and properties that are supposed to be preserved by the mappings is not explicitly exposed. If you get the chance, I'd like to hear your opinion (as well as Michael Gruninger's, Leo's, and Professor Rector's) Thanks! ArturoSanchez: @MichaelUschold: Also, it is not clear to me how change management is represented in all these approaches/architectures. That is to say, when the ontologies change, what happens? ArturoSanchez: @MichaelUschold and the rest of the team: to document architectural concerns, I would suggest to use the notations proposed by the Software Engineering Institute (CMU). "Documenting Software Architectures", latest edition (Addison-Wesley) PatBarkman: thanks Micheal AmandaVizedom: Thank you Leo! AntoinetteArsic: bye Leo ToddSchneider: Leo, there are apps that allow the generation of source code from an ontology (e.g., Top Quadrant) PeterYim: MichaelGruninger is presenting ... on slide#2 now ArturoSanchez: @ToddSchneider: it would be interesting to analyze--or come up with--software development frameworks that support the lifecyle of ontologies, including automatic code generation injected into the application and directly derived from the ontology/ies. ToddSchneider: Instead of using application as the distinguishing criteria it may be more effective to use problem space. SteveRay: @Todd: I agree that if the application is used for categorization, then we at least need a mapping provided from problem categories to application categories. anonymous morphed into Peter Bahnsen ToddSchneider: Arturo, I've been referring to this as ontologically driven development (with the subtext of displacing UML/SysML). SteveRay: @Todd: ....keeping in mind that we want to provide information to people trying to make the case to stakeholders that know what problems they are facing. i.e. I'm agreeing with you. ArturoSanchez: @ToddSchneider: I think MichaelGruninger's characterization is with respect to the functionality that can be implemented by "ontology-based" or "ontology-driven" applications ArturoSanchez: @ToddSchneider: ... and therefore, the characterization is domain-independent ... ToddSchneider: Problems resolved by applications realized by functionality SteveRay: @Arturo: I think Todd's point is that if I were trying to make the case to someone, say in my company, I might know what problem I'm trying to solve, but may not be sure what functionality would be best to invoke. Or even if I knew what functionality was appropriate, my deciding vice president might not recognize it in those terms. ToddSchneider: @Steve, exactly. Referring to what John Sowa suggested, you have to convince someone they have a problem (that your solution can solve). SteveRay: We could provide guidance such as: Problem category --> Needed functionality --> Ontological approach --> Benefits ToddSchneider: @Steve, sounds good and make a simple graphic. ArturoSanchez: @Todd & Steve: yes, your points are well taken, but that is why the summit has diffferent tracks. This track is about "A Framework for Understanding and Classifying Ontology Applications". There are other tracks that address the concern you are expressing. Now, it would also make sense to document the architecture of well-known exemplars of ontology-driven/based applications, for specific domains, which CIOs and CEOs associated with these domains can relate to. AntoinetteArsic: It would be awesome to have instances of the ontologies for these in KM and Decision Support. SteveRay: @Arturo: Agreed. ToddSchneider: @Arturo, I think 'Framework' is the important part, 'application' may not be the best qualifier/criteria for categorization (relative to the larger goals of the summit). PeterYim: = Alan Rector presenting now = ArturoSanchez: Very good presentations ... sorry I must leave now. I'll catch up through the Wiki. PeterYim: @Peter Bahnsen - Hello, would you let us know which organization are you affiliated with, please? Peter Bahnsen: @ PeterYim I'm contracting with the GSA and am here on invite of Susan Turnbull PeterYim: @Peter Bahnsen - welcome! we are glad you can join us today PeterYim: @SusanTurnbull - Hi Susan! Peter Bahnsen: Thanks for the welcome! I am new to the ontology community PeterYim: @Peter Bahnsen - hope you like what you see here! Antoinette Arsic: on slide 5 PeterYim: @Antoinette - thank you MichaelGruninger: I agree with Arturo -- the purpose of this track is to provide the frame of reference for discussing how ontologies are used and how to demonstrate benefits. AmandaVizedom: @Steve and all: I think we do need a multi-dimensional categorization. As suggested, there are things we'd normally think of as "application"-oriented, such as the functional requirements and technology context. There are also user characterizations, subject matter considerations (single-domain/cross domain), process considerations, organizational goals (things down the road for which ontos should be reusable? mandates? legal, policy, other issues?), and on. MichaelGruninger: If we tell someone that ontologies provide the technology that they need to solve their problems, we need to ensure that they understand exactly what is being delivered. SteveRay: @Michael: Agreed, but someone should provide a mapping from problem categories to application categories. Seems like that would either be this track, or possibly Track 4 - Strategies for making the case. MichaelGruninger: @Steve: What are examples of problem categories? PeterYim: @AmandaVizedom, MichaelGruninger and All - "multi-dimensional" is the key challenge ... how can we effectively document and present it SteveRay: @Michael: Let's see... "We don't know how to assign the right expert to a customer account", or "I'm trying to optimize the load balancing in my electrical grid". These are off the top of my head. Some answers may be obvious, some not. Kind of gets into systems design. SteveRay: @Michael: I just realized you asked for categories, not examples. I'll have to think further. AmandaVizedom: I don't think is news as an idea, but I've had a great opportunity to experience and observe the factors recently. I've been working on Ontology Design Pattern & Best Practice documentation for ontology developers within our project, translating general ontology best practices and ODPs to specifically what to do in our context. Because target audience is (a) often only lightly trained, learning on the job, and (b) not necessarily interested in more general level, except where understanding it will help them again later, and (c) mixed in desire/ability/tendency to find related external resources and try to understand and integrate them, I need to be able to articulate, in complementary guidance, in what ways these docs are tailored. Meantime, I'm thinking about how the tailored docs could be part of a larger body of guidance docs in which context one could find the bits that apply to ones own cases. I've not gotten to the point of listing such factors (not part of client task, of course), but the variety is very salient. YuLin: @MichaelGruninger: I don't think gene sequencing is using ontology technology. Could you please give an example? BruceBray: sequence ontology is an evolving example of use of ontology for representing gene sequences see: http://www.sequenceontology.org/ MichaelGruninger: @Yu Lin: I was referring to ontologies (such as the Sequence Ontology) that are used to support the analysis of gene sequence data e.g. query all databases for all genes whose transcripts are edited, or trans-spliced, or are bound by a particular protein. YuLin: @Bruce Thank you very much. I think it rather a tool for data integration than for the sequencing analyzing. YuLin: Thank you Michael, I got what you meant ToddSchneider: Thank you to all the speakers. Have to go. AldenDima: @PeterYim,@SteveRay - For me, a high-level way of presenting the categories revolves around saying that ontologies enable efficient 1) coordination 2) cooperation and 3) coherence between sofware systems and their users. I'm sure that there are other high-level categories as well. FabianNeuhaus: @Alan: about integrity constraints: it might be of interest to you that there is a W3C Member Submission called "Validating Semantic Web Data with OWL Integrity Constraints" on its way AlanRector: Glad to hear it - I have been concerned that this stream was critical but not getting the attention we require. RamSriram: @MIchaelGruninger: I believe the track's title is "Application Framework." I presume you are going to synthesize the presentations into a "framework" perspective. Such a framework should help us to resolve various views for ontologies and applications. AlanRector: One important difference is uses of ontologies as payload or "coding" or a common structured vocabulary - e.g. the Gene Ontology or SNOMED - and use of ontologies as structure for software or standards. PatBarkman: Search Engine Optimization. Were looking for a way to use an ontology for SEO because we have a dataset that blends and makes really no distinction between a broad selection of knowledge domains (for example: biology, physics, cosmology ...) In the vast number of data sources these subjects are considered separate disciplines -- segmented. Our dataset defines commonalities and treats everything as one subject. Well, you might think that sounds great ... but it kinda sucks for SEO because our web content is semantically very different than the rest of the web so we score poorly based on the existing Search Engine algorithms. So, were looking at creating Topic Pages based on our data. Then when those Topic Pages get crawled we believe we will begin to educate Search Engines about the commonalties between various disciplines and that will enhance our search engine scores. Then we also need to make our data (which is in English) searchable in multiple languages. And frankly, we also need an ontology to search our data more effectively ourselves. PatBarkman: so, those are our use cases: SEO, multi-lingual & enhancing our own search capabilities of a dataset that's likely unprecedented in it's cross-disciplinarily nature PatBarkman: wouldn't the use cases be the set of system requirements for developing the ap framework?? SteveRay: @Pat: What is your dataset? Could you expand on this? PatBarkman: @steve our data is in the form of text, video & audio, discusses all aspects of science as one subject PatBarkman: ...kinda hard to describe other than as a "complete" dataset SteveRay: @Pat: Could you say what purpose your are trying to fulfill? Education? Consumer service provision? .... PatBarkman: OK ... thanks for that clarification ... Ap FW summit/community PatBarkman: @Steve - education, primarily SteveRay: @Pat: So perhaps educational curriculum material might be an ontology. In this forum we have talked about an ontology for curricula (last year in fact). PatBarkman: @Steve, so I can probably find something in the archived discussion threads on that, right? PatBarkman: thanks SteveRay: Yes, I'll try to dig up some links. JimDisbrow: The Ontology Case Study I tried to put together (on an Energy-Water-Climate nexus) has failed so far. The lack of funding may be the single largest factor. The lack of volunteers (and not getting done what was needed) was also a factor. PeterYim: @AmandaVizedom - please capture you point here ... it's great! AmandaVizedom: Here's the point I just made on the call: In addition to looking at and collecting use cases, there is tremendous value in looking at them comparatively and *specifically* looking at what worked and didn't work in each case. Doing this kind of collection and comparison of LL in many use cases is also valuable, as discussed a bit last year, for identifying Best Practices and building that body of shared knowledge we don't yet have. But comparing what worked and didn't in different use cases is often just the light one needs to bring out the dimensions (and important points along those dimensions) we're looking to understand. Comparing use cases with LL, we can ask "What are the differences between those cases, such that one thing worked here and another thing worked there?" AmandaVizedom: In follow-up, Michael, Alan, and others noted the difficulty in getting people to talk about their failures, and that these haven't been solicited as part of the summit use case call. Michael added this to that call. I agree, and having been aiming to get a workshop set up at one of the major conferences to talk about use cases and LL, collaborating to identify both meaningful differences between the use cases and best practices. PeterYim: Great session ... fantastic presentations from all the panelists! Thanks you All! Bye! PatBarkman: thanks! AlanRector: Thanks bye AntoinetteArsic: Thank you bye SteveRay: Logging out. Thanks Michael and team for another great session. PeterYim: I will keep the chat board going until 11:30am PST (i.e. for another 8 minutes or so) ... past that, what goes into the chat-board will not get captured into the session proceedings. PeterYim: - session ended 11:21am PST --