A method for building displacement in automated map generalisation

The automation of the map design process through map generalisation continues to be a challenging area of research. It is acknowledged that a diverse range of techniques are applied during the process of map generalisation and these have been mirrored by the creation of a range of algorithms that mimic these discrete operations (such as typification, aggregation, selection). This paper discusses in detail one such algorithm that resolves conflict among objects through displacement. Perhaps more critical than the algorithm itself, is the stage prior to the application of displacement (identification, modelling), and the phase after application (the evaluation). It is argued that these two stages are absolutely critical to the successful design of automated systems. The paper begins with a review of other approaches to displacement and then describes a methodology that encompasses detection, resolution through displacement, and evaluation. This methodology has been implemented in Stratege, an object oriented expert system devoted to contextual generalisation. Details of the implementation are given, and results are illustrated using 'real' geographical data. The results are evaluated and the applicability of the entire methodology is discussed in the broader context of other map generalisation algorithms, based on the explicit representation of constraints.