The analysis/design dialectic works by factoring away, from a complex "problem domain", sub-domains that can be handled in relative isolation.
Many criteria may be used to identify commonalities, and different people will apply them in different order, with different priorities.
Let's imagine a model where any identified node defines a
"location" (hopefully offering a scope). The various locations may be
"placed" with respect to each other.
One is likely to produce a graph,
Did I depart already from the most general case to follow my
personal inclination? This process is the one Aristotle followed
first. It is highly "discriminative" as nodes are identified in as
much as they differ from each other.
It may depend on the criteria one has chosen. Aristotle would have
selected criteria based on meaning, he would have favoured "essential"
--as opposed to "accidental"-- criteria, and the resulting graph would
embody the underlying semantics.
Orthogonality, remaining semantics?...
Mutual exclusion: versions