Re: [ontolog-forum] Foundation ontology, CYC, and Mapping

[Ali Hashemi <ali.hashemi+ontolog@xxxxxxxxxxx>]

Dear John and Mills,

Here's a summary of a baseline for a further exploration of the issues surrounding such a repository, adapted from (Hashemi 2008) - go to http://reseed.ca/ali it is the 3rd file, entitled: " Hashemi - Ontology Repository as a Community.pdf " ) or follow this link: http://reseed.ca/Ali/docs/Hashemi%20-%20Ontology%20Repository%20as%20a%20Community.pdf

======================
John,

I'm glad we're in accord. I think it is important to separate the repository component of such a project from the services that wrap around it. I agree that people should be able to use whatever notation / syntax / dialect they prefer though have a requisite that before inclusion in the repository, they be mapped to CL.

We really want a curated repository as well, where we can have a "wild west" section, and another where the contents have been verified. Setting up a mechanism to address these concerns is important. Points 3-6 that you raise seem to fall more in the services category (mapping translating external representations in, providing tools for browsing etc.). While very important, I think it would help to delineate the tasks that these objectives generate.

It will be useful to discuss these issues at the summit. Though many of these points have also been already raised in earlier meetings of OOR.
======================

Mills,

See comments inline.

On Wed, Mar 3, 2010 at 11:30 AM, Mills Davis <lmd@xxxxxxxxxxxxxx> wrote:

I'm encouraged by the direction of this discussion thread. Especially, I think that your advocacy of a lattice or hierarchy of ontologies is a good one if we can agree on conventions for instantiating sibling ontologies, ontologies that are more general, ones that more specialized, as well as rules for identity and provenance of versions. It seems that these same conventions should apply well when someone maps two or more ontologies together creating a knowledge fabric, which is a new instance that should have its own identity and provenance.

The group at the Semantic Technologies Laboratory is developing a DTD for capturing just such these issues. They provide an extension to the vocabulary developed by the the Ontology Metadata Vocabulary (OMV) project. Specifically, we are providing terminology to capture exactly the scenarios you describe. If you haven't already, take a listen to the OOR talk dated Feb 19, 2010 ( http://ontolog.cim3.net/cgi-bin/wiki.pl?OOR/ConferenceCall_2010_02_19 ).  At least in Michael's presentation, slide 4 shows five of these relations.

An ontology can extend another one Conservatively or Nonconservatively.  Two ontologies may be related to one another via Relative Interpretation or Faithful Interpretation or be Definably Equivalent.
At the moment, since OOR does not have the adequate infrastructure to support these relations (or really, first order theories), these will be captured at the metalevel.

What this means for someone who wishes to include an ontology in the repository is that they need to keep track, at least of the first two relations. The next three arise from how their ontologies interact with others.

As an example for sibling ontologies, you could write an ontology for lattices uses a single relation leq or two functions, meet and join. Let's call them LaT1 and LaT2. In the repository, these two distinct ontologies are related to one another by (DefinablyEquivalent LatT1 LatT2).  This is stored in the metadata. Incidentally, two other ontologies, RCC-mereotopology and Stone-lattices are also definably equivalent, so we would also have (DefEquiv RCC Stone).

Now what about the actual mapping axioms? They reside in a distinct module, depending on which way you would like to map (if they are not symmetric) there might be one or two such modules. Since this "knowledge fabric" is a distinct theory/ontology/module (see http://ontolog.cim3.net/cgi-bin/wiki.pl?FormalOntology ), it too would have metadata relations culled from the above.

There are more relations pertaining to how primitives are propagated through extensions, how new terms are defined, whether a term is a definition extension or not. I'll see if Michael can share this information as it would save the group a lot of effort. Of course, a service will have to be developed to automatically run through these links and generate the types of views that John mentioned above.

Some overall observations -- while it would be useful to characterize all possible relations and metadata, this is unlikely to be done any time soon. My inclination is that we should aim to define a minimal set of rules so that we can start inputting ontologies sooner than later.

I suspect the hardest part at first will be education and formalizing / adapting existing ontologies.

My challenge to the Ontolog community is to come together to provide some practical guidance for how (global) communities such as this one can best approach the management of networked data, data fabrics, and data sharing.

Agreed!

Ali

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (01)