Ordnance Survey – Great Britain's national mapping agency
![Ordnance Survey home page [logo]](/images/gui/logo-os.gif){kind=logo}
6.3: Spatial databases (3)
Object databases
Object-oriented databases have their origins in the realm of object-oriented programming languages. This subject is notoriously difficult to explain in simple terms and it may be advisable to seek more detailed resources and devote some time to fully grasp the concepts of object-orientism.
OODBMS organise information very differently than RDBMS. Rather than spreading the information about an entity across a range of linked tables, it is stored together in discrete lumps called objects. Each object is defined within a hierarchy of object classes so that it inherits properties from a parent class. Additional attributes can be defined within the object and they are said to exhibit encapsulation because they can self-describe their own particular set of properties, and therefore the way in which they can be queried. Object-oriented databases can make it easier to model real-world phenomena in a logical form.
There is a particular problem for the GIS student learning about object databases: the use of the term object. The situation gets clouded by the fact that geospatial data corresponds to real-world objects. In GIS the shapes and locations of things are stored as coordinate geometry. GIS data is often stored in a database, either storing the coordinates as numbers or using special geometry data types. You will hear GIS practitioners refer to an object database to describe any database that can store the geometry of topographic objects in its tables. To a practitioner of pure computer science the distinction between relational and object databases has nothing to do with geography. To make things worse, there is a hybrid type of database called object-relational in which advanced data types can be stored in relational tables that reproduce some of the advantages of the object model. In theory, you can store object geometry in each of these database types: relational, object and object-relational.
As you can imagine, it is very important when using this jargon that you know exactly what you mean by an object! The term spatial database is better for specifying the storage of geographical features.