Re: "factoring out" in the construction of FuGO = the search for "universals"

[William Bug]

Please see below for inline comments. Cheers, Bill On Jul 8, 2006, at 8:06 AM, Xiaoshu Wang wrote: To use the example you have chosen, for an anatomist to want  to use FuGO and be ontological compatible, they need only use  an ontology founded on the same definition of continuants &  occurents.  How does this differ from what I think you are  claiming.  An anatomist need not necessarily care about the  details of FuGO entities bearing no relation to data they  intend to semantically integrate with (such as  'LC_instrument') to remain fundamentally compatible with  FuGO.  They only need agree on the definition and use of the  foundational levels of the ontology they share in common - in  this case 'Continuant', 'Independent_ continuant', and  'Object'.  In fact, this would be encouraged by the OBO  Foundry participants - use FuGO for formal semantic  descriptions of assay/instrument/reagent provenance and use  the Foundational Model of Anatomy to deal with anatomy (right  now, just human anatomy, but that is being worked on) - both  of which are founded on BFO. I think we have discussed the problem of "web closure", see Chimezie's wiki here http://esw.w3.org/topic/HCLS/WebClosureSocialConvention.  We have to realize that you cannot selectively import a subgraph from an RDF model. And using owl:import, in effect, tightly bind one ontology to another.  If ontological terms are not carefully grouped under different namespaces, we will eventually end up a one-for-all schema, which is neither practical nor useful.   This is a big concern for what we are trying to do at BIRN - and - as I mentioned - what is being proposed under the OBO Foundry principles (http://obofoundry.org/).  Both of these are very large, field-wide efforts with an increasing amount of resources being invested, so it will be important to make these issues clear to the community of biomed. ontology informaticists currently using OWL for these purposes. I ran into exactly the problems you describe, when trying to selectively export branches of our BIRNLex semantic graph.  I tried applying separate namespaces to make this work.  I really do understand the technical/engineering issue you are discussing.  I just wasn't able to make this work under Protégé-OWL, either because: 1) I'm ignorant of the tools available to support this task; 2) The tools don't exist (tools for adding namespace qualifications to subgraphs in an OWL file do exist in Protégé, but I wasn't able to use them to accomplish this task) 3) As others on this list - or in the TCon have suggested - the complexity and interdependence of nodes across disparate portions of underlying graph really prohibits one from doing this. The "modularization" you refer to via namespaces relates to a  conceptually different issue, as I see it.  Even though the  use of namespace - especially in XML & RDF - has advanced to  something considerably beyond its original application of  simply avoiding tag/term/name "collisions", the  "modularization" namespaces can support in XML space is  distinctly different from the "factoring out" of universals  to which Trish referred. When I think of "factoring out", I thought of software engineer's term of "refactoring" to decouple the hard-coded dependency.  Yes, when I replied to Trish, I want to state explicitly what I mean by "factoring out". This was clear in your post.  The point I was making, is Trish is referring to a distinctly different process derived from the OBO Foundry Principles for ontology construction. Using the namespace facility inherent in XML formalisms might  be a useful way of enabling people to exchange different  world views for sub-regions of the graph, that's true.  But  it is by no means a requirement.  One should still expect  whether or not you are using namespaces to chop up or  modularize different portions of the ontological graph, you  are still following the same principle of extrapolating to  universals, even if the universals for "namespace_a"'s  coverage of the cardinal parts of a liquid chromatographic  apparatus differed from the universals defined for  "namespace_b"s coverage of the cardinal parts of a liquid  chromatographic apparatus . I think you misunderstood the topic.  The factoring out to "universals" etc. is semantic issue.  For instance, what makes a "is_a" relationship, or Guiono's OntoClean Methodology.  I don't think I misunderstood, as I mention above The factoring out that I mean, a.k.a., modulization is an engineer issue.  The task is not difficult to envision, imaging that you are writing an RDF reasoner, and given one assertion from say ":foo a fugo:continuant" Can you provide a more specific example.  In the formalism being applied in the FuGO curation process, the only nodes which would map to ':foo' this N3 triplet would be ':independent_continuant' and ':dependent_continuant', and I can't think of any instance data that will map to those nodes in the graph. , try to play around it and see how many assertions you will be end up with.  For human beings, there is no such "engineer" issue.  When we speak one word, we never intend to pull out the entire dictionary.  But semantic web is dealling with machines.   I really do understand the point is to support algorithmic processing of formal semantic statements - from determining approximate semantic equivalence of entities across separate Knowledge Maps up to more advance reasoning applications.  I also have a very deep appreciation for the problems that can arise when one conflates the lexicon with an ontology - similar to the distinction Nicola Guarino makes between "a set of extensional relations describing a particular state of affairs" and the "intensional relations" or "conceptual grid which we superimpose to various possible state of affairs" (http://www.formalontology.it/section_4.htm). If namespaces were used as the primary means of "factoring  out" universals, you could end up with quite a proliferation  of namespaces for FuGO - in the extreme - one for every node  in the graph. Yes, of course. It is a price we have to pay in order to maximize the sharing and reuse of an ontology.  Nothing comes for free. Of course, how to modulize is more of an art than science.  But there are certain principles that we can follow.  One is Gruber's "Principle of minimal ontological commitment" (Gruber, Toward Principles for the Design of Ontologies Used for Knowledge Sharing, Knowledge Systems Laboratory. Stanford University., 1993).  The other that I proposed the Principles of Orthogonal Domain in my manuscripts.    Orthoganality is one amongst the several OBO Foundry Principles.  I will look to the proposal in your manuscript, but can you quickly take a look at the OBO Foundry Principles and explain how your proposal differs from what they propose.  This is a very important issue, because these principles are being applied now across a very broad field of biomedical ontology development projects.  If you scale that to the many instance data repositories for which the ontologies are being - and will continue to be - used to create knowledge maps/association files, one can see a very large amount of the data space HCLS semantic web projects will want to tap into will be fashioned according to these principles. I wish I could understand what you describe at  http://www.charlestoncore.org/ont/2005/08/o3.htm a little  better.  I found your Nature Biotech. paper very clear and  extremely helpful in several discussions I've been having on  the issue of XML vs. RDF, but I really am having a hard time  with your description of O3 on that web page.  Is there a  publication where you go into this in more detail? Yes, I did have a manuscript written and send to Journal of Applied Ontologies recently.  I tried to keep the informaiton on the web to be minimal because I don't want to incur any potential conflict.  But I will send you my manuscript in private. Many thanks, Xiaoshu. Xiaoshu Bill Bug Senior Analyst/Ontological Engineer Laboratory for Bioimaging  & Anatomical Informatics www.neuroterrain.org Department of Neurobiology & Anatomy Drexel University College of Medicine 2900 Queen Lane Philadelphia, PA    19129 215 991 8430 (ph) 610 457 0443 (mobile) 215 843 9367 (fax) Please Note: I now have a new email - [EMAIL PROTECTED] This email and any accompanying attachments are confidential. This information is intended solely for the use of the individual to whom it is addressed. Any review, disclosure, copying, distribution, or use of this email communication by others is strictly prohibited. If you are not the intended recipient please notify us immediately by returning this message to the sender and delete all copies. Thank you for your cooperation.