The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
The later describes the intension knowledge (TBox) about 
geographic concepts. Both of them constitute the knowledge 
base (<TBox, ABox>) of a particular domain or task setting, so 
that different contents can be easily matched when integrated. 
Based on the proposed framework, explicit semantics could be 
formalized and embedded on the top of general representation, 
which will increase the usability of geospatial resources, 
furthermore, facilitate sharing and exchange through service 
composition in cooperative computing environment. The 
implementation of geospatial data semantics and geospatial 
function semantics will be briefly discussed in the following 
subsections. 
3.2 Geospatial Data Semantics 
The geospatial data semantics refers to the meanings or 
interpretations of geographical data in the view of information 
representation. In the proposed framework, geographic concepts 
extending from geoFeature constitute the geo-data concept 
lattice, which usually describes the static structural domain 
knowledge. 
Figure 2 shows a section of geo-data ontology about chemical 
hazards. Some geographic concepts, for example, 
ChemicalFacility, ThreatArea, as well as the relations and 
properties, are described by OWL. In distributed collaborative 
environment, different ontologies can be developed according 
as levels of abstraction. Then domain knowledge can be shared 
and reused through concepts extension (is-a) and aggregation 
(part-whole) within or across ontologies. In this example, both 
POI and AOI are defined in another ontology, GeoOnto, and 
they are all derived from geoFeature. 
<owl:Class rdf:about-'"«‘SpecialLocations"-' 
<nlfs;subC lassOf rdf:resource="^Dt>mai nC oncepti.’'.^ 
<rdfs:subClassüf rdf':resource= "http:, www.tsgis.nju.edu.cn' 
Opettgt‘i£rGeoOntologies/GeolMo.owt#POr> 
</owl;Class> 
<Owl:Class rdf: ID="ChcmicalFaci lity"> 
<rdfs: comment rdf:datatype= , 'http://www.w3,org/2001/ 
XMLSehematfstring^faeilmes producing chemical gas 
</rdfs:eommem> 
<ndts:subClassOf rdf:resource=*’ttSpecialLocations",'> 
</ovvl:ClaiiS> 
<owl :C lass rdf: about="#ThreatArea"> 
<rdfs:comment rilf:dalatypc="http:.vwww.w3.orgf'2001 
XMLSchema^string">dangerous area caused by toxic gas dispersion 
</rdfs:coinment> 
<rdts;subClassOf rdf:rcsource="^Sccnanos'V> 
<rdfs:subClassOf rdf:resource i ="http:/www.lsgts.nju.edu cn/Opengtser/ 
GcoOnlologiesGwOnto.ovd# AQrv> 
</owl:Class> 
Figure 2. Geospatial data semantic representation based on 
geoFeature 
3.3 Geospatial Function Semantics 
Geographic concepts extending from geoOperation constitute 
the geo-task concept lattice, which describes the dynamic, task- 
related knowledge of application domain. From service-oriented 
perspective, every geoOperation can be viewed as a function of 
geo-data, with input or output parameters wrapped as 
geoFeatures. It is designed to formalize the user actions in 
geospatial space. The semantic relations between actions or 
geoOperation can be divided into four different relationships 
with entailments: troponymy, Proper inclusion, backward 
presupposition and Causation [Kuhn, 2001]. Similar to 
METEOR-S [Akkiraju, Farrell, Miller, et al., 2005], some 
semantic tags are added to web service description language 
(WSDL) in our work to provide explicit semantic description of 
service interface, including input, output parameters and the 
constraints of operations. So, GIServices described by this 
model would be easily discovered and composed with the help 
of consistent semantics. 
Figure 3 shows a simple example of geospatial function 
semantic representation. In this example a geoOperation 
concept, identify Threat Area, is defined to formalize an action in 
geospatial problem-solving environment. The input and output 
parameters of geoOperation refer to geo-data ontologies for 
capturing explicit semantics. And some constraints to execute 
the operation are also defined by two tags, precondition and 
effect. 
¡¡coopération name " AppOnt: Identi fy Threat Area" * 
• input message "DtnOtit:ChemicaiFaeility'7> JL _ 
output message "DmOftt:ThreatArea'7> ■ 
outputs 
• Precondition “ AppOnt#Preeimd Expr" > 
- effect "AppOnt#eftect Expr" > 
geoOperation > 
geoFeature 
dvcmicalFaciltly 
* 
threaiArea 
radius 
thrcatArea 
Ideo tifyThreat Area 
\ 
geoOperation 
IdcntifvthrcatArca 
Figure 3. Geospatial function semantic representation based on 
geoOperation 
4. CASE STUDY: COLLABORATIVE EMERGENCY 
SERVICES 
4.1 Application Scenario 
This example is motivated by the need to support geospatial 
information sharing and collaboration among emergency virtual 
organizations. GIS and geospatial information are indispensable 
in all stages of emergency management, involving immediate 
response, recovery, mitigation and preparedness. Collaborative 
emergency management requires multiple individuals and 
organizations sharing information, expertise, and resources in 
support of rapid situation assessment and decision-making. It 
relies upon geospatial information to depict geographical 
distribution of events, its cause, affected people and 
infrastructure, and available resources. 
Here we consider a simplified evacuation scenario of toxic gas 
dispersion, posed by the related GIServices and collaboration 
through formal semantics. When a toxic gas leakage detected in 
some chemical facility, many social departments cope with