Evaluating and Refining Computational Models of
Spatial Relations Through Cross-Linguistic Human-Subjects
Testing
David Mark, David Comas,
Max Egenhofer, Scott Freundschuh, Mike Gould, and
Juan Nunes COSIT `95, Semmering, Austria,
A. Frank and W. Kuhn (eds.), Lecture Notes in Computer Science, Vol. 988, Springer-Verlag, pp. 553-568, September 1995.
Abstract
Human subjects testing based on a sound formal model is outlined
as an effective way to evaluate and refine computational models of
spatial relations. The cognitive response of a subject (person) to
a real-world situation depends on the characteristic of the
situation and the characteristics of the subject. Spatial entities
can be characterized by topological relations, metric, scale
(scope), kinds of phenomena (semantics), motion (time), and (for
maps and displays) graphic presentation. SubjectsÕ responses
may vary according to their natural language, their culture, and
individual characteristics such as gender, age, and handedness.
Given this conceptual background, a battery of seven human subjects
testing protocols are introduced. Each test is described, and
results of applications of these tests are summarized where
possible. The tests are compared with their requirements, and the
different aspects of human spatial cognition that they might test.
Lastly, a program for applying these tests and refining the formal
models based on test results is presented.
PDF
David Mark, David Comas,
Max Egenhofer, 
Scott Freundschuh, 
Mike Gould, and
Juan Nunes