Database Reference
In-Depth Information
Although tempting, particularly to the GI Department within Merea Maps, it decides
not to do so as it feels the boundary will be an artifice that could be challenged
depending on the decisions it makes to define the boundary. Rather, it realizes
that what is important to most people is not a boundary, but the buildings and the
access points to these buildings; people want to know which buildings belong to
the Hospital, what the buildings do, and how to access the Hospital. Merea Maps
therefore constructs a mereological description of the Hospital, not a geometric one.
Every Entrance is a Feature.
Every Hospital is a kind of Place.
Every Hospital is intended for
Health Care.
Every Hospital has part a Building
that is intended for Health Care.
Every Hospital has part an
Entrance.
Class: Entrance
SubClassOf: Feature
Class: Hospital
SubClassOf: Place, isIntendedFor
some HealthCare, hasPart some
(Building that isIntendedFor
some HealthCare), hasPart some
Entrance
By describing the Hospital as a Place, logic dictates that the hospital has a
Footprint related to some geometry. As the actual extent is unknown, this Footprint
could either be a point feature
14
that relates to a representative position, perhaps cor-
responding to the main hospital building (if there is one), or it could comprise all the
polygons of all the buildings that make up the Hospital. Both of these are imperfect
solutions; there is no perfect solution.
As we have seen, there are of course places that do not have well-defined boundar-
ies. Frequently, localities within towns and cities are examples of these. Quite often,
such places were once separate and distinct villages but over time grew together or
were absorbed by a growing town or city and lost their administrative independence.
As time passes, the boundaries between such localities begin to blur, so the centers
are well defined but the peripheries begin to overlap, with people having differing
opinions of what lies within one locality and what lies within another. Geographic
Information Systems (GIS) have difficulties handling this, although a number of
techniques have been developed or adopted that provide approximate solutions, such
as the egg yolk method (Cohn and Gotts, 1996) or kernel density (Rosenblatt, 1956).
However, both these methods rely on some degree of quantification of the vague-
ness at any point and so cannot be elegantly represented as Linked Data or easily
described ontologically. As with the Hospital example, a mereological model of each
area could be constructed. This would represent an area based on the buildings,
streets, and outdoor areas, such as gardens, parks, and recreation grounds, that are
known to be a part of the area. However, whereas in the Hospital example Merea
Maps knew with certainty which buildings belonged to the hospital, in the locality
example, we only know with certainty those Features near the center of each locality.
Merea Maps handles this uncertainty by introducing a subproperty of “is part of,”
namely, “may be part of,” and allows more than one locality to use this subproperty
on the same feature as shown in
Figure 10.8
. Here, buildings B9-B11 and B14 are
considered by some to be part of Altby and by others West Chine. A classic exam-
ple of where such differences occur can be seen by examining the different views
that may be presented by estate agents who are motivated by the need to present
