The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
2. METHODOLOGY 
Two crucial technical issues are involved in our work: 
geographical semantic formalism, and geographical semantic 
sharing to enable service-oriented Geo-Collaboration [Cai, 2005; 
MacEachren, Cai, Sharma, et al., 2005] in distributed 
computing environment. We explored combining Ontology and 
Web Services technologies to provide an integrated semantic- 
based framework for collaborative GIS application, mainly by 
ontology-driven semantic formalism and sharing geographical 
semantics under Web Services architecture. 
2.1 Semantic Representation Based on Ontologies 
Semantic research in GIS community has to tackle some major 
issues concerning the characteristics of geographic concepts, 
which implicates the task of establishing a geographic ontology. 
Generally, the design of geographic ontologies should take into 
consideration the fact that geographic entities interrelate, 
participate in processes, present variation of properties and 
values, etc. The task of defining geographic concepts, 
determining relations among them, and finally establishing 
axioms, necessitates a comprehensive examination of the 
semantics of all that constitutes geographic space. Using current 
web-based Ontology description language, such as OWL 
[Harmelen, Hendler, Horrocks, et al., 2004], domain knowledge 
can be logically formalized and shared through internet among 
different development groups that work on different application 
tasks of the same domain. 
Geographic concepts are an integral part of geographic 
ontology as they stand for mental representations of all possible 
things (real or abstract) that exist or may exist. Therefore, the 
core problem of geographical semantic representation is the 
formalization of geographic concepts, which form the backbone 
of knowledge base about the domain or task setting. Current 
effort of geographic ontologies design pays more attentions on 
the static and structural domain knowledge, containing a view 
of the world that has less to do with human activities than 
existing data holdings. These, in turn, are usually based on map 
contents rather than on an analysis of actual user needs. In order 
to enhance geo-information exchange at different levels of 
semantic granularity, two important elementary geographic 
concepts, geoFeature and geoOperation, are advanced in our 
approach to describe basic data semantics and behaviour 
semantics separately in GIS application domain. Moreover, 
both geoFeature and geoOperation are not strictly limited to 
describe spatial concepts, which will help to deal with 
geospatial and other resources uniformly. 
2.2 Semantic Sharing with GIServices 
GIS has been evolving from GISystem to GIServices rapidly 
[Gong, Jia, Chen, et al., 2004]. To share geospatial data through 
internet, the Open Geospatial Consortium (OGC) has developed 
a series of web-based interoperability standards and protocols, 
for example, Web Coverage Services (WCS), Web Feature 
Services (WFS), Web Map Services (WMS) and Web 
Processing Services (WPS). They allow seamless access to 
geospatial data in a distribute environment, regardless of the 
format, projection, resolution, and the archive location. Now 
OGC web services are widely used in geospatial communities 
and are forming basis of the new-style WebGIS applications. 
One important obstacle to integrate GIServices in distributed 
computing environment is that different systems use different 
concepts and terms for describing service interface. In our work, 
explicit semantics of interface elements, especially input or 
output parameters and constraints of geoOperations, are 
formalized and embedded into GIService description. 
Geographic ontologies are designed to define the vocabulary of 
unambiguous domain related concepts, and meanings of the 
geographic concepts anchored in consensus domain knowledge. 
This common understanding of the terms and concepts that 
describe given domain is especially important in distributed, 
collaborative computing environment where the clients may be 
geographically distant from each other and working on 
heterogeneous software platforms and programming 
environments. 
3. GEOGRAPHICAL SEMANTIC REPRESENTATION 
FRAMEWORK 
3.1 General Framework 
According the above ideas and methodologies, an integrated 
semantic representation framework is given based on service 
oriented architecture in this section (Figure 1). 
Class 
GIService 
Class 
Geol'eMure 
SubClassOf 
dm 
ontology 
Class 
GeoOperation 
SubClassOf / 
X 1 
—— —f— 
task 
r ontology 
Y 
4 
\ 
\ 
< 
InstaneeOf 
, s 
-■ GeoOper»tion> 
• 
- 
<Output> 
Object level 
Instances of GeoFeatures Instances of GeoOperations 
1 
SDBI 
Source level Distributed GIS Database Geo-processing Resources £ 
Figure 1. Service-oriented geographical semantic 
representation framework 
The framework is composed of two distinct, yet interrelated, 
components: the Data component and the Knowledge 
component. The Data component represents the raw and 
uninterpreted geospatial resource, which can be organized into 
spatial database or files geographically distributed. The 
Knowledge component, defined by DL-based ontologies, 
represents the logical semantic knowledge derived from 
uninterpreted geospatial resource. Furthermore, geographical 
knowledge can be described at two levels of abstraction: object 
level and concept level. The former describes the extension 
knowledge (ABox) about instances and how they interrelate. 
1086