The current conceptual model for data representation in spatial information systems (SISystems) has been challenged by various problems. In this thesis, a new theory of representation is presented based on the study on how spatial features are represented in the human mind in order to meet those representation requirements. It follows the idea of a spatial feature based representation. Spatial features are defined as the natural units of substances in the world and the basic building blocks of the environment.
Specifically, the biological principle is followed in understanding the representation of spatial features in mind. A situation based Ontology of the world will be introduced to provide a figurative re-interpretation of Piaget’s operative account of cognitive development. That is, based on this idea of situations and Piaget’s theory of cognitive development, four levels of cognitive systems and four stages of conceptual systems are distinguished. Thus, Piaget’s abstract structure d’ ensemble can be re-defined with part-whole relations on the one hand, a system of domains or a system of ontological-kinds of spatial features on the other hand. Regarding the representation of spatial features, conceptions are explicitly differentiated into four inclusive levels for the states, individuals, specific categories, and the ontological-kinds. Four kinds of causal relations are also emphasised as the constraints for the development of domain specific representation and knowledge. With the new theory, constraints of core-domains on representation in cognitive literature in terms of ontological-kinds of identities and causal knowledge of different kinds will be explained in an unified account. The constraint of ontological-kinds on representation of spatial features can be differentiated in terms of whether the ontological distinctions can be supported or not and what forms of representation can be given. The factors on different models of spatial features can be understood in terms of ontological-kinds, basic level categories of core-domains, details of category systems (of different kinds) of core-domains, and forms from interpretations of attributes of different kinds.
This understanding of the structural constraints on the representation of identities is ultimately for developing a new information architecture of SISystems to adaptively handle the different models for spatial features (of different ontological natures) and the representation of the environmental situations in general (including maps). The emphasis of the constraints on the representation of spatial features necessarily separates this thesis from other representation theories in SIScience which focus on the forms of representation with attributes directly.
