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

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

PatC,

Three more things,

1) Somewhat more constructively -- if you are looking to create an FO, i think current "core-hierarchies" might also be another good place to explore in the search for primitives. Each core hierarchy acts in effect as an "FO" if you like, for a particular domain. So when I speak of mutually inconsistent ontologies located within a family of ontologies (core hierarchy), i am speaking of divergent views that may share a common parent. Each core hierarchy is an unbounded partial ordering (it's not necessarily a semi lattice, because there may not always be a unique join).


2) You have disparaged the results I showed you as being not practical or too restricted or, to be honest, I don't understand your criticism. The results are extensible. Maybe you don't see the implications of these results?


Perhaps you might point me to a "practical" large scale FOL ontology that is actually being deployed? I'm willing to wager that a large part of what is said in any such ontology is reducible to concepts from mathematics or other generic ontologies (i.e. temporal etc.). Indeed, i'd say this is almost guaranteed, since the ontology is represented in a formal language, likely using commonly understood logical structures. Anyway, I'm really curious and would be fascinated if there existed many such ontologies that did not map into others in the way we've described. Indeed, it would be an astonishing result! ... ...


3) The fact that meanings are interconnected does not mean that modularization is a fantasy. I suggest you pay closer attention to the ways meanings can be connected. How does the addition of Geometrical relations to Time theories change the meanings in both? I'm not sure how familiar you are with the notion of conservative and noncoservative extension. Look them up. It makes the problem of "interconnectedness" a lot less daunting than you think. Take a look at the modularization of PSL. All the meanings are interconnected, yet they somehow made it modular. And it certainly is _not_ a toy ontology.


Cheers
Ali

On Thu, Mar 4, 2010 at 11:25 AM, Ali Hashemi <ali.hashemi+ontolog@xxxxxxxxxxx> wrote:

Pat,

Three overall comments.

1)  You asked me for a case where metadata linking novel ontologies to those already in the repository didn't exist. This is akin to mapping in a novel ontology that is not already linked to the FO. _That_ is the apt comparison, not one where the target ontology is already integrated. If that is what you are assuming, then allow me to assume that metadata mappings into the repository exist :P. The mapping problem doesn't magically disappear....

2) you wrote:

[PC] you indicate that the modules may be inconsistent with each other.   That can create big problems in trying to use such a set of modules.   A mapping or integration algorithm may be able to detect inconsistency, and perhaps there will be some consistent combination of modules that will actually satisfy the needs of some realistic application.  Unless a serious effort is made to make modules consistent (the kind of effort that would be made for the FO), I expect that finding a set of modules that satisfies a need and is consistent would become exponentially harder as the number of inconsistent groups of modules increase.  And ultimately, I would expect that finding a suitable set of consistent modules will be much less likely than finding a finite number of semantic primitives.

I'm not sure the intent of "inconsistent modules located within a family of ontologies (say about temporal ontologies)" was effectively communicated if this is the conclusion you draw.  Inconsistency between two modules within this family is not a detraction nor does it imply the system breaks. It simply reflects reality. Different people working on different problems will characterize problems according to their needs and partial views. This characterization may or may not be consistent with another characterization.  The repository keeps track of which modules are inconsistent with one another (it's in the DTD).

Additionally, having these modules in the repository means that you can trace the hierarchy and find the point at which the modules are not inconsistent and use that as the basis of agreement. Mutual inconsistency is not necessarily a bad thing. Two lower level modules being inconsistent doesn't mean that elsewhere within the family of ontologies, there is a more general module that is consistent with both. In fact, this is the whole point of core-hierarchies (families of ontologies) !

3) I doubt anyone is going to wait for large scale consensus on global primitives before creating an ontology.  If (and before) that goal is ever attained, many new ontologies will have been developed, not to mention the existing ones. Moreover, if a new ontology is made without plugging it into the FO - which i suspect is highly likely, then i could, the same way you bandy the ungrounded charge that :

"I seriously doubt that anything but the tiniest fraction of applications will be relatable by that tactic alone"

I could replace *that* with *FO* and throw it right back at you.  If someone does not use the FO, or has an ontology that predates mass global consensus on an FO, then you have the exact same mapping problem we're trying to address. Let's compare apples to apples if we're making comparisons, shall we?

Cheers,
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)