The OWL ontology for each MEANING class cites required aspects
The OWL ontology for each property limits its legal values

By putting MEANINGS into classes, rather than unnamed concepts, we
can control how they are modeled, and exploit some prototype images.

• lexikos/concepts:  Roget category models should be here, in RDF, as PSIs
• they can have POS data + CAT data to support inheritance
• they can have INGLISH models built partly into SHAPES
• we can EXTEND these conceptually with associated XTM prototypes
• each is a set of descriptors that expand the core concept
• the set gets inherited, unless other descriptors override it
• lexikos/agent/corelex:  XTM, used here, can define an (extension) list of words in context
• a shared baseline list can be pre-built from our Rogets-data
• it can MERGE into an Lextend extension list in each context
• the CORELEX data used - plus syntax - can be the combo
• lexikos/agent/discourse: these refer to instances of messages & shapes in the text
• this (XTM?) defines the linguistic data stream itself, the speaker(s), etc
• this (topic map?) is needed to define "what is said", by whom, etc.
• each instance of a word & phrase is a "layer 2" (RDF?) object
• each NP includes a pointer to some "layer 3" topic(s) in context
• clause models can become XTM associations that remain in discourse
• their structures directly make up "layer 4", then get "interpreted"
• these VP structures may/always? also become topics in "layer 3"
• lexikos/agent/context: images modeling *real* instances of Things being discussed
• typically, the image comes from the FIRST topic occurance in discourse
• that image cites RDF model of one individual Thing, using concept props
• need not have a URI, unless it has been deemed a topic?
• typically, it will be marked as an instance of some concept
• at layer #3, each individual Thing can instantiate multiple concepts?
• these, not their shapes, are what we must model as 'topics'
• any associations used here refer to reality models, not words

By storing images in RDF, we can exploit the storage features, yet keep
selected OWL features at arms length!

I'd like to add a simple syntax for scribbling the above (but defer this)
By omitting this, one can only define new properties/classes in the ontology?
However, WORDS allows new aspects/kinds to be based on the above?

Defining meanings of terms:

To work with prototypes, we define a 'kind' as a special class instance
To define meanings for terms - declare most as a kind (a class prototype)

(term ako -parent classes-) => a meaning for the term denoting the kind
        (ako -parent kinds-) => image of a new kind (per first parent-kind)

Each kind holds a list of (default) descriptors that agree with class spec 
Each listed descriptor looks roughly like this:  (aspect -values-)
Each aspect is the defined meaning for some OWL property

Defining meanings and images by using the above:

(term isa -parent classes-) => meaning is an instance (prop name)
        (isa -parent classes-) => image of an instance (per first parent-class)

(term aso -parent classes-) => meaning is a set of instances (prop name)
        (aso -parent classes-) => image of a set of instances (per plural, first parent-class)

Modeling changes to the default descriptors

the list of new descriptiors can be replaced below by only the apsect -values-
 
(put -object- -new descriptors-) => shape (updated) or complaint, per -prop-
(-object- has -new descriptors-) => shape (updated) or complaint, per -prop-

Describing objects:

(-definition-
  (prop1 -values-)
  ..
  (propN -values-))