The Development of Phenomenological Generalization Within an Object-oriented Paradigm

Abstract

This paper presents a "phenomenological" framework that incorporates information relating to 1) an object's geometry, 2) its semantic properties, and 3) the inter-object relationships. It is argued that this information is required in order to devise algorithms and design strategies capable of abstracting the meaning inherent in the phenomenon being mapped. The fundamental challenge is to design generic algorithms that are powerful and flexible enough to provide varying solutions according to this information. It is argued that this can be achieved by implementing functionality at the object level, and that the object-oriented paradigm provides a flexible modeling environment that obviates the need to develop bespoke algorithms for each and every feature type. This approach is applied to extend a geometric-based algorithm for area feature generalization into a phenomena-based algorithm which takes into account the semantic meaning and contextual relationships of the real world objects. Solutions derived from the algorithm demonstrate its success at generating varying results for different types of phenomena and map subjects.