Multi-Criteria Geographic Analysis for Automated Cartographic Generalization

Cartographic generalisation is a process similar to text summarization that transforms a map when scale is reduced. Cartographic generalisation simplifies the map content while preserving as much as possible the initial characteristics and spatial relations of the map. The automation of this process requires a deep understanding of the context of each map feature, which involves different criteria such as the shape of the feature, the semantic nature of the feature, or the spatial patterns of its neighbouring features. This is why multiple criteria decision techniques can be relevant make multiple criteria decision during the process. This paper proposes two use cases of cartographic generalisation sub-tasks where multiple criteria decision techniques improve past techniques: the classification of urban building blocks and the ordering of the buildings to keep in the map as a priority. In both use cases, the proposed methods are experimented on large real cartographic datasets, and evaluated in comparison to alternative techniques.