‚BASE 
al database 
. There are 
th the batch 
s of errors 
abase. This 
ital data. In 
we define a 
or}. Spatial 
object? In 
le model of 
Knowledge 
th change, 
, however, 
systematic 
€ no given 
sh is the 
cription of 
of updated 
y provide 
st transfer. 
ographical 
each user 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
Typically, the whole new versions of a data set are currently 
disseminated to end users even if only 10% of the objects stored 
in the database have changed (this is a common estimation of the 
rate of evolution per year for geographic databases (Raynal, 1996). 
By bulk transfer, the users either ignore the update (if it is not 
significant enough) or manually (and selectively) update their 
database and rebuild their value-added data. In some cases, users 
can’t accomplish their tasks. So, we have to adopt new transfer 
mode. Incremental Data Transfer is the preferable solution that is 
a data transfer system that enables suppliers to supply only 
information that has been affected by change between two 
versions of a dataset and for users to be able to incorporate those 
changes into their data. Many researchers proposed various 
spatio-temporal data models describing the geographical 
changes(Peuquet, 1994; Frank, 1994; Langran, 1992; Worboys, 
1994; Jiangjie, 2000). But up to now, no single spatio-temporal 
model in a GIS has been adopted. Therefore, a description and 
classification of geographical changes is required. The analysis of 
geographical changes has been one of the important directions of 
the GIS research agenda over the past ten years (Peuquet, 1994). 
Several classifications of geographical changes have been 
proposed, changes at different levels of abstraction (Hornsby and 
Egenhofer, 1997) and the description of spatio-temporal processes 
that generate these changes (Claramunt and Theriault 1995, 1996; 
Hornsby and Egenhofer, 2000). And these researches describe a 
geographical change with the comparison of two pictures that 
denote the two states of the objects. This way implicitly contains 
people's cognitive process, that is to say, this description has 
limited information about geographical changes. 
The objective of this research is the identification of a model for 
the classification and description of geographical changes. We 
explicitly describe people's cognitive process about real world 
changes and express these changes as a set of rules. A 
geographical entity is represented as a four-element group 
{semantic component, thematic component, temporal component, 
and spatial component}. Based on this model, we propose 
taxonomy of geographical changes. 
3. REPRESENTATION OF GEOGRAPHICAL OBJECT 
Geographical objects are always presented as a 3-tuple {thematic 
attribute, geometric attribute, ID}. This model defines semantics 
by thematic attributes. Semantics in this paper is defined as the 
relationship among the computer representations and the 
corresponding real world feature within a certain context (Bishr , 
1998). The major problem of defining semantics by attributes, is 
to identify which attributes to choose. In general, it is extremely 
difficult and often impossible to find a complete set of 
characteristic attributes for a Real World object (Kuhn, 1995). In 
order to transfer more information about geographical object and 
its change to users, we prefer a model with more semantics to 
represent the geographical object. We replace the ID with 
semantics description and add temporal description in the 
classical representation of geographical object , ie. a 
4-tuple{ semantic descriptor, thematic descriptor, spatial 
descriptor, and temporal descriptor }. Spatial descriptor consists 
of geometric component, position component and topologic 
component . 
4. CLASSIFICATION AND REPRESENTATION OF 
SPATIO-TEMPORAL CHANGES 
4.1 The definition of spatio-temporal change 
What's the change? Let's see some examples of changes, spread 
of forest fire, building collapse by earthquake, alternation of land 
access, and so on. Object is the database representation of real 
feature. By oriented-object method, we can represent the real 
feature as the object. Object is the basic semantic unit for us to 
understand real world. The definition spatio-temporal change is a 
process of the transformation from one state to another state of the 
geographical features. Spatio-temporal change influences one or 
more geographical features and their attributes. So, there are 
several factors related to change, object set, object, object 
component and component attribute. We give a 4-tuple to 
represent the composition of a change {ObjectSetb, ObjectSeta, 
name of a change, rule set }, ObjectSetb represents object set 
before a change, ObjectSeta represents object set after a change, 
rule set contains many rules by which we use to identify the 
change. 
4.2 Capturing semantics 
Capturing semantics of change is the identifying process of a 
change. The identification of a real world change is a heuristic 
process. Heuristics are common sense knowledge, or rules of 
thumb, that originate from the expert's past experience (Avelino 
et al., 1993). There are three categories of heuristic knowledge: 
associational, motor skills, and theoretical. Associational 
knowledge is mostly acquired through observation. It is always 
represented as rules( for instance IF—THEN). Hence, 
representing such knowledge as a set of rules is a viable solution. 
Data, geometric and thematic, and knowledge can be 
encapsulated into an abstract object type with sufficient discipline 
599