Conceptual Aspects: How Landmarks Can Be Described in Data Models - Landmarks: GIScience for Intelligent Services - page 108

spatial representation of an individual. Even in Turing's original definition of the
intelligent machine it was the machine's behavior rather than the machine's (always
finite) knowledge base that enabled the machine to pass the Turing test.
Secondly, even if completeness is not needed, an extensional approach of storing
(finite) lists of objects that may serve as landmarks is insufficient. This must be the
only valid conclusion from our insights on context-dependency of landmarks.
Thirdly, spatial intelligence cannot be based on what current spatial databases
'know' about the world already. Spatial databases contain vast numbers of entities
over a range of types, and are certainly not limited to store further entities. This
means, features representing the objects that are used as landmarks in mental
spatial representations are either already in spatial databases or can be added easily.
Actually, from an ontological perspective, the superset of all features in spatial
databases is the universe of discourse for identifying features of landmarkness.
These features represent the objects human conceptualization has found worth
recognizing, for whatever reason. The machine also anchors each of these features
to their location in its own spatial reference frame, such as the World Geodetic
System 1 [ 37 ] . Furthermore, it is in the nature of spatial databases that these entities
are collections of shared concepts. (Some) people have agreed on the taxonomy
as well as the individual entities. The entities have frequently names allocated by
some authority, such as geographic names, addresses, or institutional names. But the
linking of these features to the landmarkness of the represented objects is missing.
Even the current spatial databases' taxonomies do not contain a geographic concept
landmark . Why is that? After all what we have heard in the previous chapters
it has become clear that landmarks are not a geographic category comparable
to building or river —notwithstanding that even these 'tangible' geographic types
are hard to specify formally in their semantics due to conceptual vagueness and
cultural dependency. Instead landmarks do not appear to be a geographic concept
next to others. We have learned that landmarks are rather a cognitive concept, an
internalized embodied experience statically linked to a location. Thus, a landmark
is a role that an object—any object in geographic space, from geographic scale to
table-top scale—can take. Practically this makes a landmark not a (class) type in a
spatial database, but a property of the entities in a spatial database.
Before defining an agenda, it may be useful to consider whether some functions
of an intelligent system can be realized already, and what the limitations are.
￿For understanding a human-generated message current state of knowledge can
parse the message, the locative expressions (references to objects) can be
isolated, and location references both by type (“the library”) or by instance
name (“State Library”) can be resolved by matching with database content [ 44 ] .
There are limits, though. A reference to an object type will match with a
1 https://en.wikipedia.org/wiki/World_Geodetic_System , last visited 3/1/2014.