Ontology-Driven Generalization of Cartographic Representations by Aggregation and Dimensional Collapse

Automatic generalization of cartographic features has been recognized as a goal of Geographic Information Science (GIScience). Many successful algorithms have been introduced for generalization tasks such as point reduction and smoothing of linear features. Such algorithms operate well as a function of change in map scale or resolution. Other generalization tasks have proved considerably more difficult. Two of these operations, aggregation and dimensional collapse, are trivial to implement --- replacing a set of points with an area feature or replacing an area feature with a single point --- but have proven challenging to make operational. The decision to aggregate or collapse features is as much dependent on the context of the features as they are change in map scale. This dissertation proposes to show how ontologies can be used to inform automated generalization in these operations.