Spatial Inference Based on Geometric Proportional Analogies

This paper addresses the domain of qualitative spatial reasoning, addressing two distinct domains of spatial information: namely geometric proportional analogies (GPA) and topographic (land-cover) maps. We present an analogy-based model that unites two distinct models that were developed, one for GPAs and the other for topographic maps. We describe how this new generalized model solves GPAs and also solves some inference tasks associated with topographic maps. We then describe how the new model makes use of additional categories of spatial information such as point features, which have received comparatively little attention. Finally, we present a new category of problem related to the identification of irregular structures, which lay beyond the scope of previous models but which is solved by our newer model.