Conceptual Aspects: How Landmarks Can Be Described in Data Models - Landmarks: GIScience for Intelligent Services - page 115

would be as much a wasted effort as it is for the speaker to consciously weigh
between all available landmarks in their mind. In this situation there is no optimal
description. Many of them (ideally all of them) would help me finding the way,
and thus, pragmatically they would have the same information content. Some of
them may even have the same smallest number of references, satisfying brevity.
Still their references may be different. One may relate a turn to the location of the
Post Office, while the other prefers the fast food restaurant of global brand at the
same intersection. If both would be successful for the wayfinder, then clearly none is
better than the other. Numerical differences in landmarkness do not matter as long as
the task can be successfully completed. Choosing landmarks is a matter of sufficing
rather than optimizing; and since this is the case for a human speaker, it can be the
case for the intelligent machine as well. In summary, we will complete our formal
model by making landmarkness context-dependent, and by acknowledging that an
object of some landmarkness in a given context is sufficient for communication in
this context.
Modelling context in depth goes far beyond the scope of this topic, and modelling
context comprehensively is even impossible. Context could be modelled in a
hierarchic taxonomy, for example, starting at top level with the distinction of
spatial tasks made by Sadalla et al. : (a) landmarks in routing tasks, (b) landmarks
in orienting tasks, and (c) landmarks in memorizing tasks, e.g., for places or
events. Each of these categories could further be split, for example, landmarks
in routing tasks will be different for different modes of travelling (because of
different embodied experiences of the environment). Further levels of context
specifiers can be introduced: the particular individual with their personal mental
spatial representation, or the set of people this individual could be assigned to
by their shared degree of familiarity with the environment (e.g., first time visitor
versus local expert), or the language and cultural background of the communication
partner (e.g., familiarity with medieval city outlays versus familiarity with new
world grid outlays), or the time of the day (e.g., daylight versus lights at nighttime).
As mentioned, the list (or hierarchy) is open-ended.
Just to illustrate how landmarkness can be modelled in a context-dependent
way, let us set up a simplistic model that may even violate hierarchical groupings
deliberately:
data Cid = Pedestrian | Motorist | Tom
- context identifier for a number of predefined contexts
In Haskell, the vertical bar means an exclusive disjunction. The context identifier
can take one of these alternative values, and thus, can describe landmarkness of
an object either for a pedestrian, or for a motorist, or for Tom, an individual. This
simplistic model will allow at least a context-dependent storage and reasoning about
landmarkness. A smarter machine would replace this model by a model with more
elements or more (hierarchical) structure.