Introduction:What Landmarks Are, and Why They Are Important - Landmarks: GIScience for Intelligent Services - page 23

If we go back to the human-generated response, we may have now a better
appreciation of the feat the human mind has accomplished in understanding the
request. From the conversation context people guess that 'airport' most likely
refers to Melbourne's international airport, Tullamarine, and they may open with
a confirmative question (“Tullamarine, you mean?”). The response conforms to
Grice's conversational maxims [ 20 ] , especially to the maxims of relevance and of
quantity (“Make your contribution as informative as required,” and “Don't make
your contribution more informative than is required”). It also applies some other
remarkable principles. For example it mixes conversational modes between verbal
description and pointing (“this street”), engaging with the user in an embodied
manner. It also avoids to rely on quantities (such as in “after 493 m turn right”)
by using a qualitative and deliberately uncertain description (“at the hospital turn
right”). A remaining uncertainty about distance is resolved by the reference to a
landmark, the hospital. Finally, it folds all further instructions along the highway
into one (spatial chunking), relying on knowledge in the world—the signage to
the airport—thus avoiding more redundancy in the triangle between traveller, the
environment, and the speaker.

1.4 Summary

Landmarks stand out in environments, and structure mental representations of envi-
ronments in cognizing agents. They form anchors in mental spatial representations,
markers, or reference points. They are essential for any spatial reasoning, for
example, for orientation and wayfinding, and for any spatial communication. They
appear in sketches, in descriptions of meeting points or routes. People use landmarks
quite naturally.
This is in stark contrast to spatial information systems, such as car navigation
systems or mobile location-based services. In their interaction with people they lack
an ability to interpret people's references to landmarks, or to generate information
by referring to landmarks. Their internal representations of spatial environments
is certainly not based on landmarks, and thus, landmarks have to be brought in
additionally to their internal representational structure, and properly integrated such
that algorithms can add, revise, or select landmarks as needed in spatial dialogs.
Since landmarks are a concept of cognizing agents but not of systems, the
challenge addressed in this topic is to bridge the divide between the two. This
book will catch up, suggest formal models to capture the concept of landmarks
for a computer, and integrate landmarks in artificial intelligence. In order to
inspire formal modelling on foundations of cognitive science the topic has three
main parts: cognitive foundations, computational models capturing landmarks, and
communication models using landmarks for intelligent human-machine dialog.