[Top] [All Lists]

Re: [ontolog-forum] Oooh, FOL is too hard to learn.

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ali Hashemi <ali@xxxxxxxxx>
Date: Thu, 21 Oct 2010 12:56:53 -0400
Message-id: <AANLkTikD2CicQzFPNDX+sQxZwqcDgL16ss1b2UXVnHCZ@xxxxxxxxxxxxxx>
Hi Adrian,

On Mon, Oct 18, 2010 at 5:48 PM, Adrian Walker <adriandwalker@xxxxxxxxx> wrote:
HI Ali --

You wrote:

..a third way would be to get people to communicate directly with representations of models (examples), and map those into logic.

Sounds like inductive inference of logical clauses with variables from ground examples.  Lots of research in that area (e.g. Progol  [sic]), but are there any practical mapping systems?  (Mapping by hand would be hopelessly error-prone, IMHO.)

Yes there are some similarities, but also important differences too. Among these differences is the fact that communicating via representations of models can also be interactive, more intuitive and capitalize on abduction as well. Moreover, I strongly disagree that mappings would necessarily be "hopelessly error prone." Creating translations won't be easy, but it is doable, and where available can have a huge payoff. It falls along the lines of developing tools to support knowledge representation / communication etc.

John had written:

[JS] What got us into this mess is that people said "Oooh, FOL is
too hard to learn."  So everybody invented a different notation
for writing some version of it, and nobody gave any thought
to how their version related to anybody else's.
This is a remark that resonates and it seems like a missed opportunity - it would have been nice had people gone beyond trying to directly express FOL in new notations. 

I believe there would more benefit in developing "notations" for the representation of models that satisfy sets of axioms expressed in FOL (or Common Logic or some other formal language). 

The analogy I've found illuminating is to broadly compare this effort to that of skinning in software ( http://en.wikipedia.org/wiki/Skin_(computing) ).

As described in the above link, one skins a software application with a customized graphical interface. You don't create novel interfaces (notations) for the code that runs the program, but for interaction with the program. In this analogy, the software maps to an underlying ontology (or applications that are using FOL (or CL) as an underlying logical language) and skins map to different representations of models tailored to various user (subject matter expert) domains. 

Anyhow, the idea is that one creates a "skin" to represent, say graphical 3D models (based on, say a synthetic biology theory) as the primary interface for a subject matter expert (SME). The SME can then create, vet and/or modify these representations, and in so doing, indirectly but rigorously, affect an underlying, formally expressed computational ontology. Loosely speaking, this is already happening with the fUML project - think of an fUML conformant UML diagram as a skin for a CL expressed ontology of (a subset of) UML.

Though it is important to remember that the focus here is to conceptualize and create interfaces for communication via representations of translatable models and not on the logical notation underpinning the model (which is what virtually all work thus far on the topic has done...). 

Certainly, developing translations for the representation of models into complete diagrams is not easy (what is?), and ideally requires resource(s) adept both in ontological engineering and interface design, but makes the _expression_ of novel theories / ideas for a domain in such a system much more accessible. 

Moreover, it provides a nice bridge between the captured logical theory and the intended models of the user - making the problem of discussing something abstract quite a bit more concrete. Please note though, it makes little sense to use this type of "model based communication" as a silo or independent of other solutions - rather, it is more productive to think of it as existing in an ecology of communication interfaces.



(•`'·.¸(`'·.¸(•)¸.·'´)¸.·'´•) .,.,

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)

<Prev in Thread] Current Thread [Next in Thread>