Colin Aldridge, Friday 19 March 1999
Equivalence-based categorisation: an Information Science-oriented
approach to empirical knowledge.
The research presented in this seminar was motivated by the
problem of obtaining useful knowledge from multiple theme geographic data
where the size and complexity of the dataset challenges human
comprehension. A theoretical foundation for representing abstract and
empirical knowledges in databases is developed commencing with Pawlak's
(1982, 1991) theory of abstract knowledge. Pawlak's theory is founded on
notions of equivalence relations and classification. Colin Aldridge
combines these ideas with the well-known mathematical concepts of set
definition by extension and by intension to develop the concepts of
extensional knowledge (i.e facts) and intensional knowledge (e.g. rules).
The theory traverses the concepts of generalisation, specialisation,
induction, deduction, unsupervised learning and supervised learning.
Further considerations lead to proposing that empirical objects are
dependent upon, and a consequence of, an intelligent agent's sensory
experience of real-world phenomena. The a priori existence of objects in
nature is rejected. A theory of elemental empirical spatial knowledge is
developed, based on the idea that spatial experience of reality is
fundamentally dependent upon the spatial configuration of the sensors of a
sensing entity. It is shown that basic spatial relationships between
real-world properties can be represented symbolically using numbers and the
mathematical operators addition, subtraction. Empirical spatial knowledge
theory is applied to the extensional geographic knowledge represented by
chloropleth maps and their digital database equivalents.
Colin Aldridge is a former mineral exploration geologist who has recently
had his PhD thesis accepted and has joined the Department of Information
Science as a lecturer. His thesis is titled, "A Theory Of Empirical
Spatial Knowledge Supporting Rough Set Based Knowledge Discovery in
Geographic Databases". This seminar focuses on the theory of empirical
spatial knowledge that makes up the first part of the thesis.