What is
this video about?
In this introductory
video I would like to explain what knowledge representation is, how
to build and apply them. There are basically three phases involved in
the process of building a knowledge representation. Acquisition of
data (which includes staging), collation and the representation
itself.
The collation and
the representation phases of the process are mentioned here, but I
will explain them further in future videos.
You are now watching
a simulation of the acquisition phase as it collects and stores
preliminary structure from the data it encounters in terms of the
vocabulary contained within that data. Acquisition is a necessary
prerequisite for the collation phase following it, because the
information it creates from the data are used by the collation
algorithms which then transform that information into knowledge. The
statistics you are seeing tabulated are only a small subset of those
collected in a typical acquisition phase. Each of these counters are
being updated in correspondence to the recognition coming from
underlying parsers running in the background. Depending upon the
computer resources involved in the acquisition, these parsers may
even even run concurrently as is shown in this simulation.
The objects you see
moving around in the video are of two different kinds: knowledge
fields or knowledge molecules. Those nearest to
you are the field representations of the actual data being collected
called knowledge fields. They could represent an individual
symbol, punctuation, morpheme, lexeme, word, emotion, perspective, or
some other unit of information in the data. Each of them contain
their own signature – even if their value, state or other intrinsic
properties are unknown or indeterminate during the acquisition.
Those farther away
from the view are clusters of fields which have already coalesced
into groups according to shared dynamically adaptive factors such as
similarity, relation, cardinality, ordinality,...
These 'molecules'
also contain their own set of signatures and may be composed of a
mixture of fields, meta-fields and hyper-fields that are unique to
all others. The collation phase has the job of assigning these
molecules to their preliminary holarchical domains
which are then made visible in the resulting knowledge
representation. Uniqueness is preserved even if they contain common
elements with others in the domain they occupy. Clusters of
knowledge molecules and/or fields grouped together are known as
'knowledge domains', 'structural domains', 'dynamical domains' or
'resonance domains', depending upon which of their aspects is being
emphasized
https://independent.academia.edu/CareyGButler for the more information.
