The Heuristic Classification of Functional Land Use: A Knowledge-based Approach

Urban areas present an intense mixture of economic activities and extend over a vast territory. Conventionaly, land uses were classified according to a set of criteria in a dichotomous belong or does not belong perspective. However, the association of a spatial entity to a type of land use is not absolute, but to be expressed as a possibility, notably if we are investigating a spatial system at macro resolutions. The paper has two methodological parts. First, it addresses issues in the representation of knowledge on functional land use. We advocate that land use can be conceived as declarative geographical knowledge about an urban system, and we use a simple semantic network to encode it. Second, it develops an heuristic procedure of classification where a spatial entity has membership values to land uses. The assignment of a membership value applies elements of fuzzy logic through fuzzy numbers in classification rules. The procedure helps the semantic network "learn" (knowledge acquisition) about a spatial system by finding possible associations between spatial entities and land uses. Shanghai, as a complex intra-urban system, is used as a case study to test this methodology in the representation and analysis of geographical systems.