chat-transcript_20100211f_edited.txt PeterYim: . Welcome to the OntologySummit2010: Panel Session-4 - "Education & Training Quality for Ontologists" - Thu 11-Feb-2010 OntologySummit2010 Theme: "Creating the Ontologists of the Future" * Panel Session-3 Topic: "Template for Ontology Training Programs" * Co-chair: Professor BarrySmith (University at Buffalo) & Dr. FabianNeuhaus (NIST) * Panelists: o Professor DagobertSoergel (University at Buffalo) o Dr. BillAndersen (Ontology Works) o Dr. NicolaGuarino (Laboratory for Applied Ontology (ISTC-CNR), Italy) o Professor MichaelGruninger (University of Toronto) Please refer to details on the session page at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2010_02_11 . anonymous morphed into MariaKeet FrankOlken: I have joined the teleconference. anonymous morphed into ElizabethFlorescu Joel Bender morphed into JoelBender RexBrooks: I have joined the chat, too. anonymous1 morphed into JulitaBermejoAlonso RaviSharma: Barry: Would you agree that the abilities of ontologies to show relationships among data and also the ability to process data to useful information is the reason for increasing demand for ontologists recently? PavithraKenjige: We learn how to create Ontology in Software Systems Engineering curriculum (optional)... So Computer Science people can do that as well.. PeterYim: ... would the "anonymous" person please change your identity to your real name by clicking on the "Settings" button (at the top center), please RaviSharma: Barry: The three NOT categories could be closely aligned by-product specialists related to ontologists or do you believe that there is zero overlap among them and ontologists? anonymous1 morphed into JeffreySchiffel RaviSharma: Fabian: I believe you are clubbing the requirements for ontology in the use case and example module? PeterYim: Nicola: stressed "Ontological Analysis" FrankOlken: What prerequisites are we assuming? a basic course in logic? RexBrooks: @Nicola: What I think I'm hearing from Nicola is analysis of ontologies as opposed to the use of ontological principles for analysis of some domain or collection of documents. Is this correct? FrankOlken: Also, I think that in addition to logic, we should include graph theoretic approaches to ontology specification, characterization and integration. FrankOlken: I am skeptical about a 3 unit course on management and team work - I think this should be rolled into the project class. PavithraKenjige: I agree that, just understanding the principle would not give the expertise, it is the practice that makes a person an expert! But in training, one can have enough examples to create actual onotology would help! FrankOlken: Perhaps ontological analysis is intended to be the equivalent of literary criticism. We need a literature of ontologies, FrankOlken: and a literature of ontological criticism. FrankOlken: I agree with BillAndersen on the importance of integrating with existing curricula. New course approvals are very difficult. RaviSharma: Bill: What are the minimum tools and especially open source tools that would help develop the ontology applications and challenge the ontologist in the making? FrankOlken: I also agree with Bill Andersen on the importance of being able to work from existing DB schemas. AmandaVizedom: I would think that ontological analysis would include the conceptual, formal, and functional aspects of an ontology. I don't know that this is what Nicola intends, but assertion of a fundamentally interdisciplinary nature suggests this to me. What I mean, roughly: formal = logical, set-theoretic, and similar characteristics of the ontology and its content, and the implications thereof; conceptual = what is being captured? how does the ontology correspond to something someone, or some expressions, might mean; functional = for what (kinds of) uses and applications could this ontology be used? JulitaBermejoAlonso: How do you merge the ontological knowledge and skills (ontologist role) with domain knowledge (expert role)? They are not usually the same person. Sometimes, to speak "ontology-like" is not understood by experts. How is future training to address this? PavithraKenjige: Oh! Object Oriented concepts are quite powerful from a programming perspective.. if they have problem with understanding.. it is a gap one to pay attention to.. in the industry AmandaVizedom: @Nicola: how does the characterization above fit with ontological analysis as you conceive of it? PavithraKenjige: I mean one has to pay attention.. RaviSharma: Julia: That was also closely related to my question to Barry. GaryBergcross: Dagobert's comment on planning for how long an ontology will be maintained points to some need to have Ontology Management and planning covered in one of the courses. anonymous1 morphed into BillAndersen anonymous3 morphed into DagobertSoergel. RaviSharma: Amanda: Ref. your comments to Nicola - I was also thinking of analysis of quality, completeness and extent to which the original intent of ontology specification was actually met? BillAndersen: My notes will follow for introduction into the record. BillAndersen: [BA] = Bill Andersen [RK] = Ryan Kohl [JW] = Jennifer Williams [OW] = Ontology Works [BA] Introductory comments OW can best help this effort by transmitting our experience on over 50 ontology projects - more precisely software development projects that involve the use of ontology Although my heart lies with the "ontological analysis" emphasis, in connection with philosophical ontology and logic, this stuff in practice follows the Pat Winston "raisin bread" law. Philosophical material are the "raisins". You need the raisins, but they're only 5% of the total mass of the bread. The real question that needs to be answered is where "ontologists" fit in to larger software development efforts. * Closer to Info Sci than CS, Phil, Linguistics * Role in development is analogous to data modeler but broader in scope 1) [RK] Emphasis on training in standard Knowledge Representation It's a bad move to make KR an elective, but have NLP front and center as a 'core knowledge' piece. The KR component should help students understand what is hard and what is easy in their field, as well as allow them to clearly understand the connections to other CompSci and Math fields. 2) [RK] Competence in at least one modern programming/scripting language. The ability to write quick one-off scripts helps the ontologist in two ways. The most obvious benefit is a decreased reliance on a trained programmer for small tasks, such as ripping through a text file or parsing an XML file. More importantly, though, is the increased understanding of things IT that is gained from such an ability - whether the ontologist is writing models for databases or for workflow organization, the model is going to have to connect to some kind of program. Understanding those connections is a big plus for any fresh-out ontologist. 3) [RK, BA] A 3-credit course on life cycle management and team work doesn't make any sense Better to adapt existing software engineering curricula - like UMCPs CMSC435 software engineering course. The idea is to teach ontologists that they will be part of a software team so they need to understand how all that works on a real project Such a course would do the work of: 3 credits: Use cases and examples (know: 5, 6, 7, (292C) 3 credits: Life cycle management and team work (know: 4.1, skills: 2, 13) (292D) 3 credits: Project course building ontologies (know: 4; all skills) [JW] "In my experience, teamwork THEORY has not been very helpful, and teamwork in practice has fabulous complexity and is best experienced rather than discussed." 4) [BA] Data analysis and manipulation course [JW] Most projects (in OW's experience) are NOT clean slate in the sense that there are no existing models (even if the only model is implicit to an existing database structure). In addition, most have existing data (usually not or under-documented) which must be dealt with. Scripting, XML, SQL 5) [BA] To reinforce Dagobert's point - Develop criteria for integration of existing course offerings in CS, IS, LS to support curriculum. If not done then there will be risk that ontology-specific courses could not be supported long term. The upside is that the view of ontology as inter-departmental is reinforced. 6) [JW] The whole point of using ontologies is to promote model-driven application development. But there is a BIG gulf between app developers and ontologists. Ontology may indeed help in integration and such but there needs to be shared understanding between ontologists and app developers such that the latter can "code to" the ontologies. 7) [JW] No need for separate "philosophy of language" course. [BA - Language is important as Nicola points out, but] this can be incorporated into the philosophical ontology or just rolled up into a general philosophical support class. [JW] Core skills -- The inclusion of funding in the life cycle skill set is not something I have seen elsewhere, and seems to call for a set of skills applicable to marketing. I am not sure it belongs in a standard view of lifecycle management. 9) [BA] The idea of "use cases" sounds good but in practice will be difficult to achieve. Example are intelligence applications OW has done where the "use cases" are wildly complicated and would choke someone who does not already have significant experience. Best to provide some application templates as part of an ontology-oriented SWE course. It certainly doesn't make sense to make it a separate course. PeterYim: Thanks, Bill BillAndersen: :-D RaviSharma: excellent inputs Bill. BillAndersen: Tks AmandaVizedom: Working off of the comments about permanency, funding, and long-term maintenance: In many applied and/or integrated ontology settings, the idea of a permanent or unchanging ontology is considered at least practically infeasible, at worst based on one or more fundamental misunderstanding. However, that is not to say that disposable or one-off work is desired. The "missing middle," in the general body of knowledge, is how to create re-usable and maintainable ontologies *and* architectures (including structures of integreated, modular ontologies) that allow integration, extension, and adaptation with maximal fluidity and minimal rework. DagobertSoergel: Importance of ontological analysis, crystallizing the structure of a domain, developing a global view of the entities and relationships in a domain or in the body of knowledge needed to deal with a problem. DagobertSoergel: How to write axioms BillAndersen: +1 on Michael's point on teaching ATP RaviSharma: All: I think tools training is not sufficiently emphasized? any comments for tool training even if these are short or vendor sponsored courses? AmandaVizedom: If you put your hand up by mistake, you can remove it by clicking on your hand in the Queue box. DagobertSoergel: Relationships to related areas, such as building taxonomies and thesauri that are less formal but use similar techniques, use similar resources, and rrequire similar knowledge RexBrooks: @Ravi: Tools are critical. Understanding them are not necessarily easy, especially if the logic or relationships is built into the tool. This is important for choosing the tools we use. DagobertSoergel: Importance of courses where students build ontologies. Apprentice learning RaviSharma: Rex: For example these days - How do we employ people who do not have Office, Databases and HTML (Web) basic understanding, the tools training is similar as it prepares you to a methodology for ontology development or application of ontology to solve real problems whether these are search related or for mining or knowledge from information? AmandaVizedom: This thread (tools, modeling languages) reminds me: I have encountered several cases where substantial work is being done, including both project plans and individual ontologist-expert exchanges, with absolutely no awareness of the substantial research that has been done to try to enable better (more accurate, faster) elicitation and/or direct contribution to ontologies by domain experts themselves. I'd advocate that a look at the history and current state of such research should accompany learning about the tools themselves! FrankOlken: Again, I think there should some discussion of graph theoretic approaches to ontology specification, integration. See the work of Cliff Joslyn on graph theoretic approaches to ontology (taxonomy or partonomy) integration. PeterYim: @Frank - can you provide a link to that work you are recommending, please? RexBrooks: @Ravi: Exactly. I think one aspect that is coming up short in these discussions so far is problem solving. Ontological Analysis of problems is a purpose and use I think needs to have at least some focus, maybe a course or two. Apart from ontological analysis for analyzing ontologies. AmandaVizedom: Apologies; my previous is perhaps more of content point than a quality one. BillAndersen: Fabian - you're assuming stable funding. With no demonstration of value proposition (and that will be *software*), funding will dry up eventually. RexBrooks: @Amanda: I think so, too. RaviSharma: Amanda: That was the assumed prerequisite, but after or concurrent to the learning about SMEs on Ontologies and their contributiions, it would help to use actual tools and then visualize and graph or create reports etc on real problems that use ontology as the emphasis in finding a solution to a real application? BillHogan: Just a general comment: I have not seen any discussion of the distinction between applied ontologists and academic ontologists (faculty who advance the theory and science of ontology). Most of the discussion seems aimed at the former. Is training of the latter as simple as doing a PhD dissertation after getting an MS degree in ontology? BillAndersen: OW might be convinced to supply some of the templates for an engineering class FrankOlken: Pat, most of Ontology Work's applications are classified unfortunately. BillAndersen: Frank - right, but we can schematize the structure of those applications for teaching purposes RexBrooks: @Amanda: My agreement was to the comment before the short one about content v. quality. BillAndersen: Pat - many of those applications will involve proprietary or classified data and/or code BillAndersen: I think small-to-mid-size examples (as Michael G suggested) would be most useful and that can be done without exposing sensitive stuff. FrankOlken: Cliff Joslyn's papers can be found via DBLP http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/j/Joslyn:Cliff.html PavithraKenjige: Peter thanks for inviting! thanks everyone! AmandaVizedom: @Fabian Interesting point about the separation of *some* ontologists from the software. I think that's often correct. In some contexts, a corresponding distinction is made between ontology architects and ontologists in general or knowledge engineers as a complimentary speciality. That is, as either a specialization or a role within a project, some ontologists work the high-level design and architecture, including analysis of the software integration and/or human systems context. Others are focused on understanding the domain and developing ontology within that general architecture. Niether is harder/easier, IMHO, but they are quite different. RaviSharma: Amanda: I made a comment before Bil's pasted comments, please review whenever possible. AmandaVizedom: @Rex: got it. JulitaBermejoAlonso: Although Semantic Web oriented, the Knowledge Web (KW) Network of Excellence project did research on different scenarios for training (professional, masters, PhD), and the required core curricula in Ontology and Semantic Web. Might be of help? AmandaVizedom: @Ravi: Not sure I understand the question. ArturoSanchez: @FabianNeuhaus+Rest of Editors: what I am still wondering is if the strawman captures the idea of "ontologies with a purpose" in the context of a problem-solving life-cycle ... RaviSharma: Thanks everyone. MariaKeet: Julita: probably a European masters programme would be a nice outcome BillAndersen: Thanks to Barry, Fabian and of course Peter! RexBrooks: thanks all. AmandaVizedom: @Julita: definitely of use for the Requirements track! Consider posting links to Requirements Community Input page? RaviSharma: Amanda look at the earlier comment for you from me on the first part of the chat when you can. AmandaVizedom: @Ravi: Yes, I see it, but am not sure I understand it. JulitaBermejoAlonso: @Amanda: I will do so. JoelBender: Outstanding session, thank you! RaviSharma: One general observation on survey, it is emphsized that respondents be practicing ontologists, hope it does not take away valuable inputs from others in the community? PeterYim: Thanks, everyone, for the contribution ... full proceedings along with the audio recording of this session will be available after the weekend. ... Bye! AmandaVizedom: @Ravi and all: different portions of the survey emphasize input from different elements of community or different topics. RaviSharma: RaviSharma to Amanda: Ref. your comments to Nicola - I was also thinking of analysis of quality, completeness and extent to which the original intent of ontology specification was actually met? MariaKeet: thanks for organising this interesting session. bye PeterYim: - session adjourned 12:02 pm PST -