ABSTRACT
Cartographic generalisation is a two step process: an abstraction of the source data (to select or derive those required for the target product), followed by a representation process that produces a readable graphical image (a map). Research on the automation of cartographic generalisation has until recent times focused mainly on developing specific algorithms and measurements to detect, qualify and solve cartographic conflicts. The results have shown that solving basic conflicts is not sufficient to preserve the coherence of the map overall. This has led to the emergence of phenomenological studies, which focus on considering logical sets of objects rather than individual ones. Individual objects can be strongly modified (even eliminated) during generalisation, and if no phenomenological control is performed, the logical pattern of the objects can be lost, making the map inconsistent. The need of a phenomenological approach to generalisation has been discussed by various authors (Mark, 1991; Nyerges, 1991; Richardson and Müller 1991; Ormsby and Mackaness 1999). A complete discussion of geographical phenomena is presented by Hangouët (1998).
