Fundamentals of Geographic Engineering for Territorial Intelligence

International audienceintelligence, actual systems of artificial intelligence are not very efficient, essentially because of a naïverepresentation of space.As spatial knowledge corresponds to conventional geometric and topological knowledge, geographicknowledge corresponds to knowledge about geographic features in the real world even if real features canhave spatial relationships between them. In other words, spatial knowledge is based on topological,projective and distance relations; but if applied to geographic features, one must take earth rotundity andother characteristics (demography, physical geographic, economic geography, etc.) into account.After a rapid presentation of spatial relations and their properties, this chapter will detail the 12principles governing geographic knowledge. First emphasis will be given to various forms of geographyknowledge, such as located facts, geographic clusters, flows, gradients, co-location rules and topologicalconstraints.Then, based on ribbon theory, spatial relations and earth rotundity, geographic relations will be defined.For instance, let us consider two features in the real world associated with a DISJOINT relation; whendown-scaling, those objects can be associated with a TOUCHES relation. As a consequence, any reasoningmechanism must be transformed accordingly