RE: [xml-dev] Ontologies vs Schemas vs Transformations (was Re: [xml-dev] [Updated] A Categorization of Data Interoperability Problems)

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Places to start:

http://www.jfsowa.com/ontology/index.htm

http://www.jfsowa.com/ontology/toplevel.htm

and for the slow slogging but necessary math

http://www.jfsowa.com/logic/math.htm#Lattice

Summary:  The root of an upper level ontology 
is the empty set.  Every distinction spawns its 
own identity in a lattice of theories.  Neighbors 
share members but are not isomporphic.  The end 
result is a system not unlike an ecosystem of 
entities negotiating partial and time-limited agreements. 

The problem is the determination of the membership, 
the ubiquity of the means of enabling and validating 
agreement, the means of maintaining and updating 
such agreements and the costs of these given a 
resource/energy budget.  Identity of any given 
entity determined by its membership.  However, this 
leads to the problem of entities that exist in 
the ecotonal regions which therefore have an identity 
in multiple systems.

Open vs closed is a polarity of manageable properties 
that are different for different arrangements of the 
systems so designated.   A schema is a means to 
declare a closed system with the exception of the 
use of container types that insulate a new message 
or data type.  An ontology is a means to declare 
relationships but must also enable the introduction 
of new relationships so it is also closed to some 
extent.

Within themselves, systems can be said to be closed 
but within a larger deployment environment, they 
are open either in a deliberate or an ad hoc 
fashion.  Predictability is probabilistic and 
systems that require full certainty are difficult 
perhaps impossible to create mechanically.  However, 
systems that are semi-closed and are maintained 
are common.  Transformations are a means to denote 
and execute rules among and between the semi-closed 
systems.  Such rules can be written in formal logic 
or even XSLT.  This is largely a local choice given 
that any choice is not likely to be ubiquitous and 
each choice made closes the system by eliminating 
options of other systems.  Most attempts at creating 
upper level ontologies beyond the most basic categories 
fail at scale although some (see CYC) operate 
comfortably within known limits.

No size fits all.  Some sizes fit most.  The 
web will always be at its best operating at 
the stable edge of chaos.  This will preclude 
some applications from being web applications.

len