International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXVIII-4/W15 
5th International 3D Geolnfo Conference, November 3-4, 2010, Berlin, Germany 
97 
ONTOLOGICAL IMPEDANCE IN 3D SEMANTIC DATA MODELING 
Eliseo Clementini 
Dept, of Electrical and Information Engineering, University of L’Aquila, via G.Gronchi 20, 67100 L’Aquila, Italy 
eliseo.clementini@univaq.it 
Commission IV, WG IV/8 
KEY WORDS: Spatial ontology, Semantic Modeling, Geometric Modeling, Spatial Relations 
ABSTRACT: 
An impedance mismatch exists between spatial data models and spatial ontologies, between the language of geometric 
representations and the language of specific application domains. We call it ontological impedance. Overcoming ontological 
impedance is a difficult task, since various problems are involved, like the coherence between the semantic and the geometric levels, 
the abstraction from various levels of geometric detail, the aspects of knowledge representation that constitute a semantic enrichment 
of current models. Various aspects have to be integrated in the representation of concepts at the semantic level, like the spatial, the 
temporal, the functional aspects. Overcoming ontological impedance means representing and reasoning with knowledge at the 
semantic level, shifting the attention from the geometric level to its conceptual counterpart. 
1. INTRODUCTION 
Current 3D data models and analysis methods are borrowed 
from CAD/CAM applications, computer graphics, and com 
puter vision. There is no strong tradition in 3D GIS. It is un 
avoidable that existing 3D spatial data models are mostly ori 
ented to visualization (and not to spatial analysis). CityGML is 
a geographic standard adopted by OGC that is well suited to the 
3D representation of urban environments from a spatial analysis 
point of view (Kolbe, 2010). CityGML is made up of two dis 
tinct hierarchies, one semantic and one geometric, which need 
to maintain their respective coherence. Spatio-semantic coher 
ence is an important issue that needs to be enforced between the 
semantic hierarchy of classes and the geometric hierarchy 
(Stadler and Kolbe, 2007). 
In (Stadler and Kolbe, 2007), authors suggest that the introduc 
tion of spatial integrity constraints can be useful to test the cor 
rectness of geometrical representations, e.g. the fact that faces 
must be connected in the boundaries to form a volume, and, if 
they are thought at the semantic model, constraints can validate 
domain-specific aspects, e.g., a window must be inside a wall 
surface. Semantic relations now present in CityGML (Kolbe, 
2010) are the generalization (is_a relation), e.g., SecondaryRoad 
is_a Road, the aggregation (is_ part_of relation), e.g., Wall 
is_part_of Building, and the semantic/geometric link (relation 
has_type), e.g., RoofSurface has type Polygon. 
To foster semantic interoperability (Worboys and Deen, 1991), 
3D urban environments need to be modeled at the conceptual 
level, including entities, their attributes, the spatial constraints 
and rules that govern their existence, and the relations between 
entities. Generally, if conceptual entities have geometric repre 
sentations, a semantic relation between entities implies one or 
more corresponding geometric relations. 
Conceptual modeling is independent of various geometric rep 
resentations. In particular, it is independent of the dimension of 
the embedding space: e.g., modeling how cars interact with 
roads is independent from the fact that we could use a 2-D spa 
tial representation of cars and roads or a 3-D spatial representa 
tion. In fact, the conceptual (semantic) model is not restricted to 
a single data representation: therefore, multiple geometric repre 
sentations can correspond to a single semantic model. 
In this paper, we see the need of an object (entity)- based ap 
proach to spatial data modeling that takes into account the 3-D 
nature of entities from the beginning of the modeling phase. We 
push forward the need of spatio-semantic relations as the main 
semantic enrichment to be done in current models such as 
CityGML. In Section 2, we explain our motivations and in Sec 
tion 3 we illustrate an example of approach in OWL 2. Section 
4 draws short conclusions. 
2. SPATIO-SEMANTIC RELATIONS IN 
3D MODELING 
Current conceptual approaches for GIS modeling were in 
grained with the 2-D view: it was the conceptualization that 
could be considered intrinsically two-dimensional. We never 
described a road thinking to it as a 3-D object: it was always a 
2-D object (or even 1-D in most cases). Therefore, the seman 
tics of a road as a 3-D object has never been described at the 
conceptual level. We miss a conceptualization of the road as a 
3D object, while it would be very useful to describe roads as 
volumes, e.g., to describe various types of road connections or 
to model the maximum height of an underpass. 
An urgent research issue is about a remaking of all the pro 
gresses that were made on 2-D data models towards 3-D data 
models. Talking about topological relations, not only the mod 
els about topological relations about solids have received less 
attention in the literature than topological relations about re 
gions, but even the models for topological relations between 
regions need to be remade: the topological relations between 2- 
D regions are different if embedded in 3-D space. For example, 
if two lines in 2-D space are disjoint, it means that they are not 
connected, while two lines in 3-D space could be disjoint and 
still have a kind of connection (e.g., two rings forming a chain). 
There are topological properties that are simply (and obviously) 
not considered in models for topological relations in the plane. 
Some models for spatial relations that were extended or are eas 
ily extendable to 3D space are, e.g., for topological relations