18
®<
= = UV X
-
id we
ObJECT- ORIENTED | DATA. MODELS. IN. GIS
Gong Jianya Li
Department of Photogrammetr
Wuhan Technical Universit
Wuhan, P. R. China
Abstract
Currently, experts in Geographic Information S
This advanced technique, based upon the defini
operations in a modular fashion, is character
paper discusses object-oriented models for geo
with this concept has been carried
Key words: Object-oriented data model, Classification,
Inheritance, Propagation.
1. INTRODUCTION
There are three traditional data models:
hierarchical, network and relational models,
which all try to deal with the same problems with
tabular data. The relational data model is the
most popular DBMS model for GIS. For example,
ARC/INFO uses INFO, System 9 uses EMPRESS,
several GISs use Oracle, some PC-based GISs use
Dbase III. However, the relational model exhibits
some problems such as efficiency, data semantics,
model extension, object identity, and program
interface. It should be pointed out that the
three traditional data models including the
relational model do support these vital features
in various degrees.
2. THE CONCEPTS OF OBJECT-ORIENTED DATA MODEL
Object-oriented approach is a new technology in
software engineering and database. A definition
of object-orientation is that an entity of
whatever complexity and structure can be
represented by exactly one object [Egenhofer and
Frank 1989a], and every object is a single
programming entity that combines data and
procedures or functions that operate on that data
[Shafer 1988]. The object-oriented data model is
built upon the four basic concepts of
abstraction: classification, generalization,
association, and aggregation. The semantic model
is enriched by the concepts of inheritance and
propagation which describe the derivation of
properties in generalization hierarchies and
values in aggregation hierarchies, respectively.
2.1 Object
In an object-oriented approach, all conceptual
entities model objects, for example, & node, an
arc, a river or a province on a map are objects.
The object contains a set of data to describe its
behavior and a set of methods to operate itself.
That is:
Object-(ID, S, M) (1)
Where: ID is an identifier of the object;
M is a set of methods;
S is the behavior of the object, it is
represented by either attributes or identifier of
other objects. That means that an object can
include another object in an object-oriented
approach.
Deren
773
y and Remote Sensing
y of Surveying and Mapping
ystem are promoting an object-oriented approach for GISs.
tion of object types in combination with the corresponding
ized by clear coding which can be easily maintained. This
metric and attribute data in GIS. The investigation of a GIS
out and some approaches and applications are also described.
Generalization, Assoiation, Aggregation,
2.2 Classification
Classification is the mapping of several objects
(or instances) within a common class (or type)
[Egenhofer and Frank 1989a]. In the object-
oriented approach, every object is an instance of
a class. Each class characterizes the behavior of
its instances by describing the operators that
can manipulate these objects. These operations
are the only means of manipulating objects.
Classification is referred to as the instance of
relationship because the individuals are
instances of the corresponding class.
Thus, we establish an abstraction class by :
Classz(CID,CS,CM) (2)
Where: CID is an identifier of the class, i.e.
class name;
CS is the behavior of the class;
CM is a set method of the class.
Then
S€ CS and M = CM when Object Class
For example, a city GIS includes the classes
building, street, park, phone-line, etc. A single
instance, such as the building with the address
‘King street 51’, is an object of the class
building. Operations and properties are assigned
to object types, so for instance the class
building may have the properties address, holder
and building date which are specific for all
buildings. The operations like data retrieval and
update are also the same for all buildings.
2.2 Generalization and Inheritance
2.2.1 Generalization Generalization organizes
several classes of the objects with common
properties and operations to form a more general
superclass. There are two classes:
Class,7(CID4 , C84, CSp, CM, , CMg) (3)
Classg=(CID2,CSp,CSc,CMp,CMc)
Classy and Classy have the same subset CS, of
behavior and a subset CMA of method.
CS,C CS] and CSACCS9 as well as CM,C CM; and
CMAC CM9