Browse Prior Art Database

General-Purpose, Subjective, Non-Analytic Computer Interface

IP.com Disclosure Number: IPCOM000102387D
Original Publication Date: 1990-Nov-01
Included in the Prior Art Database: 2005-Mar-17
Document File: 2 page(s) / 114K

Publishing Venue

IBM

Related People

Todd, SJ: AUTHOR

Abstract

A technique for computer users of arts-based graphics to navigate in a multidimensional parameter space by means of a non-analytical interface. An interactive choice of direction of iteration towards optimization is offered to users not wishing to operate in an analytic mode. Other non-analytic users can benefit, such as a manager exploring the "what if" space generated by an economic model, or a scientist investigating multidimensional space. It is driven by subjective, non- analytic decisions, such as 'good' or 'worse'. Normal 'good' user interfaces make analytic use easier, but do not provide non-analytic use. Standard 'hill climbing' search techniques as a user interface tool are also implemented.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

General-Purpose, Subjective, Non-Analytic Computer Interface

       A technique for computer users of arts-based graphics to
navigate in a multidimensional parameter space by means of a
non-analytical interface.  An interactive choice of direction of
iteration towards optimization is offered to users not wishing to
operate in an analytic mode.  Other non-analytic users can benefit,
such as a manager exploring the "what if" space generated by an
economic model, or a scientist investigating multidimensional space.
It is driven by subjective, non- analytic decisions, such as 'good'
or 'worse'.  Normal 'good' user interfaces make analytic use easier,
but do not provide non-analytic use. Standard 'hill climbing' search
techniques as a user interface tool are also implemented.

      A task is broken into two phases, a model definition phase
performed by a 'programmer' and a parameter selection phase performed
by a 'user'.  The model definition phase is performed analytically by
conventional programming.  The resulting model takes several scalar
parameters, and produces a result that may be displayed on the
screen.  The model may also include a definition of suitable ranges
for the parameters, and an initial set of parameter values.

      The parameter selection phase is driven by a general-purpose
interface program which selects one or more sets of parameter values,
and displays the results of applying the model to them.  A menu of
subjective evaluations is offered, such as 'good', or 'this one is
best', which the user selects as he feels appropriate.  A new
'mutated' set of parameters based on the user's selection is then
generated, and the process iterates.  The user need not be aware of
exactly what the parameters of the model are or how they interact.
The mutator logic adapts its choice of 'mutated' parameters according
to previous user selections.  For example, the model may define a
color in terms of red, green and blue.  As the user tries to make a
color less saturated, the mutator logic will discover a vector in
parameter space that (locally) corresponds to saturation, and tend to
drive its choices along this vector.

      The technique is appropriate wherever a parameterized model may
be generated.  Examples are in computer art to form a sculpture, to
select color or texture, or to design custom wallpaper.  In
education, the model may be any process that is to be explained to
the student 'user'.  For business, the exploration of economic models
where parameters are managerial choices such as proportion of profit
to plough back.  In science, the exploration of a large,
multidimensional set of model parameters can give an extended viewing
transform, defining color as well as the geometric transform.  Users
who perform parameter selection may have no interest in or aptitude
for analytic thought, as when a sculptor prepares a n...