Ontologizers all,
I have previously been conceptualizing the Formal Concept
Analysis (FCA) method of categorization as the de facto standard for
classification.
But I have now been reading Lakoff's "Women, Fire and
Dangerous Things" where he proposes a different method which he claims
works for certain kinds of categories, though not others:
The cognitive models
approach to categorization is an attempt to make sense of all these
observations. It is motivated by:
- a need to
understand what kinds of prototype effects there are and what their sources are
- a need to
account for categorization not merely for physical objects, but in abstract
conceptual domains--emotions, spatial relations, social relationships,
language, etc.
- a need for
empirical study of the nature of cognitive models
- a need for
appropriate theoretical and philosophical underpinnings for prototype theory.
For example, he states that all the inputs, controls, outputs, mechanisms
and enclosed lower level activities (ICOMAs) in an action provides a category
which I have been calling a "context" of the action. But the title
of the book was chosen to show that no two of the elements of the category have
the same type, much less the same evaluations.
Note: he doesn't actually use
the words ICOMAs etc; I interpreted his statements that IDEF way. Your review
may differ.
That seems to imply that formal concept analysis (FCA), which
tallies property columns in a table of entity rows, only works on basic
level objects for categorization. Categorization of experiential things
(actions, activities, events ...) must not be done in the same way as for FCA entities,
i.e. not with FCA methods of identifying the entities.
Yet both kinds of categories, Lakoff observed by those pesky
humans that just won't behave linguistically simply.
I would prefer to consider the experiential part to be a set of
events which describe the action, activity, superordinate event, etc. Any set
of events with time stamps, IMHO, tells a story of what that concept is, does,
and all the relevant activities related to that concept would be indexed on the
events, so quickly retrievable in order. Therefore fast to operate as inverted
files.
Comments anyone? Suggestions? References?
Sincerely,
Rich Cooper,
Rich Cooper,
Chief Technology Officer,
MetaSemantics Corporation
MetaSemantics AT EnglishLogicKernel DOT com
( 9 4 9 ) 5 2 5-5 7 1 2
http://www.EnglishLogicKernel.com