Architectural Heritage Ontology
Concepts and Some Practical Issues
Politecnico di Torino – DIATI, c.so Duca degli Abruzzi, 24, 10129 Torino, Italy
Keywords: Semantics, Standard, Ontologies, 3D Model, CityGML, Interoperability, ADE, Architectural Heritage.
Abstract: Interoperability has become fundamental to the management and sharing of the data. For this reason,
international standards are published and ontologies are proposed and used for structuring databases in order
to assure information retrieval, improved analysis and correct interpretation of the data, besides the
interoperability of compliant databases. For thematic data about cultural heritage, standards vocabularies and
ontologies exits, but are not fully suitable to represent some aspects of architectural heritage. In fact, complex
spatial data have to be equally managed using these technologies, for enabling analysis empowered by the
inclusion of the spatial and geographical dimension. This could undoubtedly enrich the documentation of
architectural heritage. However, few spatial ontologies exist, which are able to correctly represent the
complexity and richness of such data. In the paper, an existing ontological model for cartographic urban
themes, OGC CityGML, is extended, in order to propose a data schema for the management of architectural
heritage multi-scale, multi-temporal and articulated data. The extended parts of the model are explained in
the paper. Moreover, some implementation aspects are considered both for the definition of the ontological
schema extension and for the management of the data using it.
The management of spatial and geographic
knowledge is becoming more and more discussed,
since informatics technologies permit new advanced
analysis and possibilities in information sharing.
Contextually, some connected principles and
concepts are highlighting new requirements for the
knowledge and new needs for the data management.
In particular, interoperability is a key issue, on which
the idea of the Semantic web, smart cities and
international standards are built (Barnaghi et al.,
2012, Chourabi et al., 2012, Schaffers et al., 2011).
A unique frame is therefore needed in order to
make the conceptualisations unambiguous. This can
be solved through the use of ontologies (Guarino,
2009, Laurini, 2015) in order to reduce the risk of
misinterpretation and possible consequent damages
or loss of information (Guizzardi, 2005). Moreover,
the definition of an explicit and shared data model
permits to produce and share open data, with all the
connected advantages (Janssen et al., 2012).
Therefore, the world of spatial knowledge
management and geographical intelligence is
developing tools for the realization of an effective
“geoweb” (Laurini, 2014). We can see the effort in
the directives of some national and international
institutions dealing with cartography or
environmental management: for example, the
INSPIRE (INfrastructure for SPatial InfoRmation in
Europe) European Directive is developed by the
European Parliament and the Council of 14
2007 (Directive 2007/2/EC)
(http://inspire.ec.europa.eu/). Equally, some
consortiums of major stakeholders and actors of the
sector are developing international industry
standards, becoming the base for interoperability and
open data. In this framework the OGC (Open
Geospatial Consortium) (www.opengeospatial.org/)
standards (among which the model for urban data
CityGML) are conceived.
standards/citygml) is an open data model, an
application schema (XSD) for GML files aimed at the
representation, storage and exchange of 3D urban
objects. The original aim (its history begins in 2007)
was to foster the reusability of 3D city models. Its
semantic definition can be equally useful to manage
the semantics of the data with the tools offered by
informatics and artificial intelligence.
Some more examples can be seen in the opposite
direction, that is, the effort of the world of semantic
thematic data to include geographic information. For
this reason, GeoSPARQL
developed by OGC as an extension of the W3C
(World Wide Web Consortium) (www.w3.org/)
SPARQL (SPARQL Protocol and RDF Query
query/), which is the designed language to query
RDF-structured data. It is considered for the inclusion
of spatial data in RDF-OWL information.
These languages are defined by the cited
organizations, and represent the crucial technology
for the application of the theories of open-data and
interoperability. Among these, markup languages
allow to write content and provide information about
which role the content plays using a both human and
machine-readable format. In particular, XML
(eXtensible Markup Language)
(www.w3.org/XML/) is used as a metalanguage for
markup: it provides a uniform framework, and tools
for the interchange of data and metadata among
applications. This is why XML is the base of most of
languages born to structure open and application-
independent data and exchange them through
application or through the web. Some of relevant
XML – based languages are for example RDF
(Resource Description Framework)
(www.w3.org/RDF/), which permits to manage
semantic data (through a triple mechanism), OGC
GML (Geographic Markup Language)
(www.opengeospatial.org/standards/gml) to archive
geographical objects, COLLADA (Collaborative
(https://it.wikipedia.org/wiki/COLLADA), which is
an interchange format for 3D models, and so on. The
structure of the XML-based files is defined in equally
XML-based formats, such as simple XML Schema
Definition (XSD), which is the one used by GML
format, or extended ones such as RDFS (RDF
Schema) - OWL (Ontology Web Language)
An example of geographic issues managed on the
web using the described technologies is GeoNames
(http://www.geonames.org/) which is a database
including toponyms gazetteers and information
related to the included named places.
Looking at the Cultural Heritage field, database
interoperability and information retrieval have
always been crucial aims for its documentation
(http://www.icomos.org/en/charters-and-texts). It is
indispensable the data to be unambiguous for
permitting correct interpretation, and to be
contextualized with metainformation.
The CIDOC (International Committee for
Documentation) conceptual reference model (CRM),
developed by the Commettee of the ICOM
(International Council of Monuments) is considered
the core ontology for Cultural Heritage (Doerr, et al.,
2007). It uses RDF-OWL for the management of
thematic data. It became standard ISO 21127.
A further existing database exploiting the
described theories and technologies are a set of
vocabularies developed by the Getty Institute
(http://vocab.getty.edu/). These are oriented to
structure Cultural Heritage related terms and items,
and are divided in four vocabularies. Art and
Architecture Thesaurus (AAT), structures
hierarchically the terms linked to the description of
the works of art and architectures. The Getty
Thesaurus of Geographic Names (TGN) differently
from GeoNames, includes also historical
denominations. The Union List of Artist Names
(ULAN) contains the names and synthetic
information about the cultural heritage authors; and
the Cultural Objects Name Authority (CONA),
describes the different denominations of a cultural
item over the time. In them the spatial component is
not present, but they can be the reference for the
denomination of parts which unequivocally have a
spatial connotation (e.g. all the architectural parts or
toponyms), or for related information (such as authors
or object names).
Recently some effort has been done also to
include geographic information in cultural heritage
descriptions. Some localisation data is tried to be
included in semantic structures: the Getty project
ARCHES (http://archesproject.org, Myers et al.,
2013), based on CIDOC-CRM structure integrates
some WebGIS function; the CRMgeo project (Doerr
et al., 2013) includes spatio-temporal representation
potentiality in CIDOC-CRM.
However, these geographic references are often
bi-dimensional and have little defined geometry
(points, lines or approximate polygons), since the aim
is not the analysis and reading of the artefact
geometry, but its localisation for a territorial reading.
Recently, another extension of the CIDOC CRM was
realized: the CRM
It is expressly realized for the
documentation of standing buildings (Ronzino et al.,
2015). However, the gap in this research could be
found in the management of complex 3D models in
connection with other parts of the city and the
landscape, which is a topic treated by CityGML.
For the particular needs of architectural heritage
information management, 2D (often small-scale) data
are not sufficient. 3D dense data have to be exploited
with higher levels of detail (high measurements and
georeferencing accuracies) and complex semantic
definition (object-oriented structures) (Laurini,
The availability of the 3D dense models is a
reached aim of survey and geomatic discipline
(Chiabrando, Spanò, 2013). However, the potentiality
of management, analysis and editing typical of
traditional GIS (Geographical Information Systems)
are at present moment reduced for this kind of data.
The development of new software structure or user
interfaces are needed, based on adapted or new
theoretical framework (Brahim et al., 2015,
Solovyov, 2012), which again permit the usability of
the systems in a real effective way.
1.1 Proposal Aims
In this research, a solution to the need of a data model
for architectural heritage 3D high-level-of-detail data
is proposed, by extending the existing structure OGC
CityGML using its ADE (application domain
CityGML was chosen as a base since it is a
standardized model already dealing with buildings in
their double dimensions: as a part of the city and as a
higher detailed 3D object. It is important to consider
this double nature also in architectural heritage
emergences, because they are often both meaningful
to the definition of the cultural values of the
considered buildings. Moreover, CityGML includes
the possibility to have multi-scale representations.
The integration of the monument in wider maps of the
city or the landscape, permits to perform strategic
analysis in a broader context.
CityGML is shared as a data model, already in a
potentially implementation-ready format. The UML
(Unified Modelling Language) diagrams are
published in the OGC encoding standard (OGC,
2012). They are already in an advanced phase of the
data modelling process, since the database design
details are specified as in a logic-level model (e.g. an
object-oriented approach is envisaged and types of
data and code-lists are defined). Moreover, the XSD
files are shared and available for the direct use for
implementation. However, for its generality in
representing urban models, it can be considered an
ontology (Métral et al., 2009, Kolbe et al., 2008). It is
in fact independent from the specific applications for
which it can be used and aims at representing a
common frame for urban 3D maps data.
Therefore, for extending CityGML including
structures for the management of spatial data
complexity of architecture and monuments, some
preliminary general reflections are reported, which
can be valid as ontological-level thinking. However,
the extension is then realized in accordance with the
formats and structures used in CityGML
(implementation-oriented), to be coherent with the
extended model and for permitting the test also in the
However, some considerations and necessities of
representation remain at present unimplemented,
about more evolved constrains to be imposed to the
In a second part, the procedure followed for the
implementation of the model is presented,
highlighting some possibilities of use of the schemes
for data archiving.
In the end, some part dealing with the kind of data
to be managed is presented, taking in consideration
the processing phases to be followed (from the
processing of the 3D model to its semantic
2 CITYGML CHADE (CULTURAL
CityGML model can be extended in order to model
further aspects linked to specific application domains.
The so-composed extensions use specific
characteristics and procedures of CityGML, being
defined as ADE (Application Domain Extension).
Some official ADEs exist
ADEs). They regard especially some urban-scale
themes, such as the noise, or the inclusive routing.
Some of these are specific on buildings, for example
GeoBIM integrates some classes derived from IFC
(Industry Foundation Classes)
Classes) standard used in BIM (Building Information
Modelling) (de Laat, van Berlo, 2011). However,
even if in the future probably the field of BIM (born
to project new buildings) will meet GML models, at
present it’s too rigid for describing Cultural Heritage
buildings, which need more flexibility.
A further research has been performed for the
extension of CityGML model in order to include
some information about the CH (cultural heritage)
nature of the building and some surface
characteristics, such as the deterioration
(Costamagna, Spanò, 2013). In the model proposed in
this paper, the characteristics of surface complexity
are tried to be included. Moreover, some attention is
drawn to the traceability of the stored information, in
order to include in the data the elements useful to
technicians for interpreting the stored information
and evaluating the degree of fuzziness of the data.
In Figure 1 the CityGML Cultural Heritage
Application Domain Extension (CHADE) for the
building module of CityGML is summarized. It is
then analysed in detail in the following subsection.
The extension has been developed and will be tested
on the building module; anyway, once its validity will
be proved, its concepts and classes can be applied also
to the other CityGML modules.
Figure 1: Synthesis of the CityGML CHADE in UML
diagram. In white the CityGML classes, in grey (black for
the whole class) the CHADE extensions and the inserted
2.1 The CHADE Components:
Research of Granularity, Flexibility
From the general to the particular, the first problem
was to include some attributes useful for the
identification of the monument and some related
information (if a CH declaration exists, what are the
related documents, who are the owners and what is
the preservation authority). Some of these have been
borrowed from previous researches (Costamagna,
Spanò, 2012), and extend the Core class
“AbstractCityObject”. It is possible to include this
kind of extensions by means of composite attributes
also in following phases, that is, the implementation
of the model, since the XSD format permits to include
complex attributes in the form of DataType,
composed by a series of further attributes. Another
interesting possibility for this case is the
“ExternalReference” class, already in CityGML,
which permits to relate the model with further
databases managing data about the same object. For
example, considering the management of the
Versailles castle, the reference can be realized to the
instance having ID:700000350 of the CONA
vocabulary of the Getty Institute which describes it
The second issue is the extension of the attribute
list for the “AbstractBuilding” class. In particular, its
function and its denomination. Both these values are
complex when regarding a historical item, since both
can change over the time, and must be archived as a
reference for researches and as an element for
understanding the history of the building. Therefore,
in implementation phase, for both a DataType is
included. The BLDG_Function data type includes at
first the function name (at present in English, specific
future works with historians could further evaluate if
considering different languages in order not to lose
meaning nuances). The reference to the URI of the
Getty Institute vocabulary AAT (Art and Architecture
Thesaurus) follows, which includes the terms linked
to the buildings function as subclasses of “single built
works by function”. The last two attributes are present
almost everywhere in the detailed added data types,
because they are of fundamental importance for
historical data connotation. The time attribute is
defined as a time object defined in the same GML
general schema. It also could be defined as a
TM_Object (time object) as stated in ISO TC211 ISO
19108:2006 Temporal Schema, but some
incompatibilities among some ISO TC211 definitions
and GML requirements persist
Language). Anyway, both schemas have issues for
detailing the time considered, as a date, as a period,
with different degree of fuzziness and with the
possibility to establish a sort of topology for temporal
data, in a temporal reference system. It is of obvious
importance for managing historical data. The second
attribute is “Source”, which is detailed, in turn, in a
data type, including metadata, reference to the source,
codes for its identification and retrieval and the same
Similarly, the attributes of the class “Room” are
extended, adding “RoomClass”, “RoomFunction”
and “RoomUsage”, all with reference to the Getty
AAT vocabulary URI. The “RoomUsage”, which can
change over the time, is detailed in a dedicated data
The, may be, more interesting part of the model is
the extension of the CityGML class
“AbstractBoundarySurface”. In the original model it
has no attributes, and can be specialized as belonging
to the main parts of the buildings (e.g. RoofSurface,
CeilingSurface, WallSurface…). The change of this
class can permit to follow with a major flexibility the
description of the parts of the buildings, which are
stratified and articulated and even small portions can
have different meanings. With “portions” also “fiat
parts” (without “bona fide”, that is defined,
boundaries) are intended. Therefore, a recursive
mereological “part of” relation is added from
AbstractBoundarySurface to the same
AbstractBoundarySurface. This permits to articulate
the surfaces in hierarchical, semantically well-
defined, multiscale and possibly topologically
defined parts. Several attributes are added and
defined following the already explained criteria.
Among these, the “LevelOfSpecialisation” (LOS)
attribute deserves an explication. The 3D models are
usually considered for the geometric accuracy, which
mainly derives from the production methods and
measurements systems. This characteristic is stored in
GML models as LOD (Level of Detail) associated to
each geometry. Even if it implies some consequence
on the level of semantic definition, it’s mainly linked
to the possibilities of representation offered by the
available data, and thus to the accuracy and data
density. The Level of Specialisation, in inserted in
order to manage the possibility to define parts and
subparts which can be recognisable on the same
model (with homogeneous accuracy and LOD) but
need to be separately specified because of the
different meaning they assume if considered in the
whole or as a singular part (Figure 2).
Figure 2: Examples of consecutive LOS specified on parts
of a homogeneous-LOD 3D model. The colours represent
the parts in which the object is divided.
A further extension of this class regards the
associated geometric levels of detail: two more LODs
are added, a LOD5, for approximately 1:50 scale and
a LOD6 for bigger ones. The associated geometry
class must be defined as a “Geometric
complex::GM_CompositeSurface”, since it is
structured and hierarchical, in the same way as the
boundary surfaces must be also semantically defined.
Moreover, it has to be related to a “Topological
Complex::TP_Complex”, deriving from the
“Topology” part of the standard ISO TC211 – ISO
19107:2003 Spatial Schema or the GML specification
(they should be harmonised for the same issues
regarding time objects). This last described part is
complex to be used with current software and
requests some more efforts. Anyway, the inclusion of
the topological relations as schematized in Figure 3
should be useful for correctly set the models.
OGC has processed some topological and
mereotopological structure for helping to correctly
store the data, but they are already oriented to linked
open data formats, without regarding GML
Figure 3: Schema of Egenhofer Topological Relations to be
included in the model and verified. Some exception can
exist (for example, a pillar can be considered only for one
half as a component of a bay, admitted that they belong to
the same classification), these must be so analysed in order
to confirm or not the validity of this model. Egenhofer
relations are considered even if they deal with 2D geometry,
since an only reference surface is considered, although
being a 3D surface.
2.2 Implementation Issues
Ontologies exploit object-oriented structures and
systems, which are unusual in current and known GIS
management systems: some of the most spread ones
(PostgreSQL-PostGIS, ArcGIS Geodatabase tool,
Oracle) implement object-relational systems, which
are hybrid systems including some constructs of
object-oriented databases, but not the whole
potentiality. Some studies about the development of
some semantic GIS have been performed, beginning
in the mid-1990s (Mennis, 2003, Fonseca et al.,
2002). In these studies, an object-oriented approach
was used as an effective solution for expressing and
storing the data meanings (Scholl, Voisard, 1992). In
this way, even more powerful systems could be built
with significant data interoperability and a reduction
of any potential ambiguity. Anyway, at present
moment few similar systems are available, preferring
to use SQL-based implementations (Belussi et al.,
2011). This is due to the necessity to adapt the
exigencies to the available platforms and software
systems and to the necessity to change the storing
methods to permit the production and management of
computationally heavy files. In next years probably
the object-oriented GIS will be developed again or,
some different interface from the GIS we know will
be improved to include spatial analysis and query
The described model has been implemented using
the method defined as best practise by OGC (van den
Brink et al., 2012). UML schemas are modified,
which use stereotypes defined by a GML UML
profile, so that their meaning can be understood by
the machine and the performed transformation can be
coherent and correct. In particular, for building the
system the proprietary commercial software Sparx
Systems – Enterprise Architect is used. Contrary to
the indications of using open source software for
managing public (and open) data, it is recommended
also in some official occasions (for example for the
management of INSPIRE schemas). The software
permits to import existing models (in this case,
obviously CityGML building module and some
general schemas such as GML are used; also ISO
19108 for temporal objects and ISO 19107 for spatial
issues could be considered). The classes, selected and
imported in the new extension model, maintain all
their characteristics and relations with the other parts
of the model they belong to. This is crucial for not to
create an isolated new model, but to be inserted in a
complex existing framework.
From this basis, new classes can be added, the
attributes can be defined and new relations can be
In particular, following the OGC best practise, for
extending an existing class with further attributes, a
subclass having the same name of the class to be
extended and stereotype “ADEElement” should be
created. The specialisation relation is marked with
stereotype “ADE”. For adding a new class, a simple
subclass having stereotype “featureType” must be
The so-formed model (Figure 4) can then be
exported in different formats, including XSD for
being used as a GML application schema. Other
interesting formats are OWL, ArcGIS workspace and
similar. From the XSD file also SQL (Simple Query
Language) relational or object-relational database
schemas could be generated, by passing through
different software, such as Altova XMLSpy, which
permits to manage XML documents. However, the
passage from an object-oriented model to a relational
one often requires some adapting transformations.
Contrary to what it seems, the passages are not so
easy, or, better, they have to be controlled and
corrected, since they need correct information or
reference files describing, in the specific software-
understandable language, how the transformation
must be done. Some specific applications and proper
UML profile exist for this aim, but they are not
always easily available and compatible in any
situation. Waiting for this progress, possibly planned
as future work, the resulting files have to be corrected
manually by editing the XML text following the rules
for the ADEs realization.
Figure 4: Synthesis of the UML model modified for
extending CityGML – Building module in the CHADE,
following the OGC best practise indication (Van den Brink
et al., 2012).
3 PREPARING THE 3D DATA
For processing and interchanging the data about the
paper case study, which is the medieval Staffarda
abbey church (in the north-west of Italy) some of the
previously described technologies are used. The main
problems along the whole workflow are often linked
to the inability of the software to manage some
functionalities and algorithms and/or formats at the
same time. For this reason, several passages in
specific software are needed.
In this research, existing and available software
are used, being not the implementation of new
applications among the objectives. In particular, open
source solutions are preferred, when possible, for
interoperability and replicability issues.
When the schemas are ready, the dense high-
level-of-detail 3D models have to be prepared. Since
the managed surfaces are complex, being composed
by miles of triangles (stored in form of rings
composing a multiple composite surface), it’s
obviously not possible to store manually the singular
points, but they have to pass through a series of
phases which permit to export them in a GML format.
We will not describe here the acquisition and
processing phases, which exploit a series of
techniques for georeferencing the model in a known
reference system (Chiabrando et al., 2013, Dabove et
al., 2014), measuring points with various methods
and variable accuracy and density (Bryan, Blake,
2000), processing the models for finally obtaining an
integrated, correct, georeferenced and optimized 3D
model (Figure 5) (Bastonero et al., 2014). We
suppose then to start the process from this point.
Figure 5: Views of the 3D model (textured mesh) of the
Staffarda abbey church, processed using LIDAR
acquisitions integrated with photogrammetric data acquired
from UAV (Unmanned Aerial Vehicle) (Bastonero et al.,
The first editing phases to be performed on the
models regard on the one hand the reduction of its
points, caring the conservation of the original
definition (Figure 6). This process can be performed
using different algorithms, not always known in
proprietary software. Anyway, the topic should be
further analysed in order to establish the methods and
the limits of this practise.
On the other hand, the model must be segmented,
so that every part must be isolated from the other for
being suitably managed geometrically and
semantically (Figure 7); moreover, the temporal
connotation must be considered in the segmentation,
since it is essential for historical objects (Donadio,
Spanò, 2015). A surface can be eventually repeated if
it is part of more than one instance having different
LOS as attributes. Also this field can offer a quantity
of techniques which should be analysed for finding
the most suitable one.
Figure 6: 3D models of the church façade before (first
image) and after (second image) reduction.
Figure 7: 3D model of one segmented capital (a segmented
part is highlighted with textured representation).
These phases can be performed in 3D model
processing and editing software, such as Hexagon 3D
Reshaper (proprietary software), which has the
advantage of managing coordinates, which have high
values such as cartographic ones, therefore
georeferenced models can be directly managed.
Moreover advanced editing tools are integrated in it.
At this point, two main alternatives are available
for translating the 3D model into a CityGML-
compliant format. The first one is the use of Safe
Software FME (again a proprietary software), which
is expressly dedicated to these operations. The second
option is the use of ESRI ArcGIS, which is equally a
proprietary software, but being widely spread, its
formats and procedures are often considered as de-
facto standards. This last one is used in this case for
this reason, even if the passage from an ESRI
shapefile format, which is based on a relational
model, doesn’t give the possibility to directly specify
the final structure of the data. Anyway, also the ESRI
processing integrates the FME algorithms in the
ArcGIS “Data Interoperability Toolbox” extension.
For using the processing integrated in ESRI
ArcGIS, some more passages are necessary. The
processed 3D model must be exported in COLLADA
format for the following transformation. This open
exchange format is not managed by some proprietary
software, therefore the model must be exported in 3D
model format (such as OBJ or PLY) and reimported
in further software able to do the exportation. For
example, the open source software MeshLab can do
that. The problem is that it has difficulties in
managing high coordinate values, so that the whole
model must be translated near the origin for this
passage. The exported COLLADA files can be then
reimported in ESRI ArcGIS, as multipatch shapefiles
(ESRI, 2008). Here they have to be retranslated to
their original position in georeferenced coordinates,
and can be exported, through the “Data
Interoperability” toolbox to a generic CityGML file.
The result is the inclusion of the geometry and the
attributes of the single parts of models in files
structured as CityGML and semantically classified as
“GenericCityObjects”. The GML file (readable as
XML structured text) has to be manually modified for
including in the description schema the CHADE and
to correctly define the semantics of each part.
In particular, the reference to the extension
namespace must be added in the heading of the file,
since there is no way of modifying the FME libraries
(used directly or through ArcGIS toolbox) for
including the extensions. Moreover, each segmented
part of the multipatch is exported as a distinct object
having a geometry attribute (in form of
gml::MultiSurface), but they are not hierarchically
structured and they haven’t a specific semantic yet
(being all “GenericCityObjects”). Therefore, the
hierarchy must be set and the correct labels must be
applied following the CityGML file format.
Moreover, all the textual attributes must be manually
filled in. In this phase it is obviously considered the
extended model CityGML+CHADE. Another
important issue is to add suitable identifiers, in order
to uniquely identify the objects for query performing
and information retrieval and for realizing some
connections, for which for example Xlink syntax
(which requires IDs for linking to specific objects) are
used. It is preferable if the IDs are composed in form
of URIs (Unique Resource Identifiers), following the
rules used in best practises also in linked data
environment (van den Brink et al., 2014). In this way,
the produced information could be more easily
translated to linked data for the effective sharing and
processing through the Semantic Web.
The Xlink syntax can also be considered and used
for the establishment of mereo - topological relations
among the parts.
XML processing softwares (some used
alternatives can be, for example, the proprietary
software ALTOVA XMLSpy or the open source
software Xpad) can validate the obtained GML file.
4 RESULTS: THE ARCHIVE IN A
At this point, the GML file could be shared through
the web and read by several applications or interfaces
for being consulted and analysed.
In this case, an open source software was used and
tested in order to read the GML archive based on
CityGML CHADE. An open source software was
chosen for two main reasons: first, for the already
cited goals of interoperability and replicability of the
procedures; secondly, because the open source
software often permit to access the source code of the
libraries they use, and possibly modify them. This is
useful in order to include the CHADE schema for the
correct interpretation of objects that refer to it.
The FZK software (http://www.iai.fzk.de/www-
extern/index.php?id=2315) was used, which is one of
the more developed available CityGML viewers. It
includes the schemas of some versions of CityGML,
and also some official CityGML ADE (e.g. the Noise
ADE). Furthermore, it has an open structure, which
can be customized by adding, for example, other
CityGML schemas to be used. For this research the
CHADE schema (in XSD) was added in the directory
of the reference files of the software for the described
data to be interpreted. This is unequivocally an
advantage of the open structure of the software.
At this point the software can read the processed
GML archive (Figure 8).
In the visualization platform the object inserted in
the archive can be read, including the relationships
among them (Figure 9), some measurements can be
directly made on the 3D model (Figure 10) and some
thematic visualization can be generated similarly to
GIS management software environments (Figure 11).
Moreover, statistics about the data are computed.
However, the application should be developed in
order to include the possibility to effectively manage
some elements introduced by the extension, for
example the inclusion of different addresses (referred
to the building but also to the owners, the authority,
etc.) gives sometimes problems in their visualization.
In the same way the links inserted in the GML file for
cross-referring the objects or for inserting references
to external resources (for example the Getty
vocabularies) don’t function in the software, because
probably some change in the reading of such
components should be done.
Equally, the levels of detail that can be visualized
are limited to the ones envisaged by CityGML. For
including the more detailed ones added in the
CHADE, the application should be modified not only
by adding the schemas but even in its tools and
Also the possible thematic visualizations are
limited to some attribute of CityGML and don’t
consider the ones introduced by the extension. The
same is for the statistics and analysis that can be
performed, which are limited to some pre-set
parameters and it would be interesting to enhance
However, these limits are connected to the
visualization platform, while the previously
structured GML file is independent from them.
Figure 8: GML model structured using the CityGML
CHADE in the FZK software interface: on the left, the
objects in the model are listed, in the centre the 3D model
is visualized and, on the right, the properties can be read.
The attributes, which are, in turn, objects themselves or data
types (and are therefore composed by a set of attributes) are
highlighted by the frames. The level of detail to be
visualized can be chosen, since the data are multi-scale (in
the left part of the toolbar, framed in the figure).
Figure 9: On the right box (“relations” window) it is
possible to select and visualize related objects (geometry
and thematic attributes). In the image, the result of the
relation of the whole object to one of its parts (the flying
buttress). They are selected in the representation and the
attributes are listed in the right part.
Figure 10: Example of direct measurements possibilities on
the 3D model: areas and distances. This can be extremely
useful for architectural heritage researchers and operators.
Figure 11: Example of thematic visualization (based on the
attribute “year of construction”).
5 DISCUSSION AND
In the framework of interoperability established by
the Semantic web theories and the world of standards,
the establishment of reference domain ontologies
become critical. Since a model for architectural
heritage lacked, several standards dealing with
building, landscape or city representation and cultural
heritage management were considered as starting
point for an extension or for their reciprocal
integration. Finally, the OGC CityGML model was
chosen as ontology for representing buildings. It is
considered an ontology for being specific application
– independent, even if implementation issues are
proposed in the published standard.
An extension of the CityGML model has been
proposed and tested in order to manage complex
and multi-scale 3D models. This considers some
important aspects of the architectural heritage both
from the spatial and thematic points of view.
The conceptual definition was implemented
using some existing tools, proposed as standard best
practise. However, some passages result still
difficult and the products need to be refined
manually by editing the resulting XML file for
obtaining a valid XSD. Future improvements will
deal with the major control and automatization of
Similar considerations can be done for the
management of the 3D models, which requires
complex steps, possibly through different software
for being prepared. In the end, a final manual
editing of the GML file is equally necessary. This
could be generally due to the closed source of
proprietary software, which does not permit to
modify the used libraries for inserting the extension
of the model. In the meantime, there is little
alternative to their use.
The fact that the editing of results is possible
using XML language is beyond doubt an advantage,
because it requires basic tools (even a simple text-
editor could be effective), on the other hand, the
required skills are not within everyone’s reach.
However, a solution is proposed for managing
the complex and multifaceted data about
architectural heritage. Important aspects are cared,
regarding the granularity of the information, its
traceability, which is essential when dealing with
historical items, the flexibility of the model, to
adapt to the representation of such unique artefacts
as monuments are, and the inclusion of thematic
data with eventual reference to external databases
The realization of standardized datasets using
ontologies permits to perform automated reasoning
on the information, in particular if shared on the
web. Furthermore, the use of ontologies enables the
interoperability of databases and the information
retrieval through the semantic web. This represents
obviously a great opportunity for research and
preservation issues, but also for management,
tourism, risk analysis and further connected
The archive at present can be used in
applications similar to the known GIS, for surfing
the archive, realizing queries, applying symbols,
measuring the model. However, also the available
platforms should be modified and improved in order
to permit a wider range of analysis and statistics and
to include enhanced visualisation options.
Future work will be aimed first at including the
real management of topology and mereo-
topological constraints in the models and in the
data, for enhancing the analysis potentialities and
transversal information retrieval.
A further improvement will affect the
connection with external reference to vocabularies
(possibly using methods similar to the use of
gazetteers for toponyms) and the inclusion or link
to further data models and ontologies: for example,
the connection to the CIDOC CRM is of primary
Moreover, the translation of the model and of
the dataset as linked open data will be essential to
better exploit the Semantic Web technologies and to
connect to similar information. This also will be a
future development of the proposal.
When these aspects will be solved, it will be a
further step towards the world-wide management of
the architectural heritage data in an effective
framework for their preservation, retrieval and
Barnaghi, P., Wang, W., Henson, C., Taylor, K., 2012.
Semantics for the Internet of Things: early progress and
back to the future. In International Journal on Semantic
Web and Information Systems (IJSWIS). 8 (1). p.1-21.
Bastonero, P., Donadio, E., Chiabrando, F., Spanò, A.,
2014. Fusion of 3D models derived from TLS and
image-based techniques for CH enhanced
documentation. In ISPRS Annals of the
photogrammetry, remote sensing and spatial
information sciences. 2. p.73-80.
Belussi, A., Liguori, F., Marca, J., Migliorini, S., Negri, M.,
Pelagatti, G., Visentini, P., 2011. Validation of
geographical datasets against spatial constraints at
conceptual level. In UDMS 2011: 28th Urban Data
Management Symposium, Delft, The Netherlands,
September 28-30, 2011. Urban Data Management
Society, OTB Research Institute for the Built
Environment, Delft University of Technology.
Brahim, L., Okba, K., Robert, L., 2015. Mathematical
framework for topological relationships between
ribbons and regions. In Journal of Visual Languages &
Computing. 26. p.66-81.
Bryan, P., Blake, B., 2000. Metric survey specifications for
English Heritage. English Heritage.
Chiabrando, F., Lingua, A., Piras M., 2013. Direct
photogrammetry using UAV: tests and first results, In:
ISPRS International Archives of the photogrammetry,
remote sensing and spatial information sciences. XL-
1/W2. ISSN: 2194-9034. p.81-86.
Chiabrando, F., Spanò, A., 2013. Points clouds generation
using TLS and dense-matching techniques. A test on
approachable accuracies of different tools. In ISPRS
Annals Of The Photogrammetry, Remote Sensing And
Spatial Information Sciences. 5. p.67-72.
Chourabi, H., Nam, T., Walker, S., Gil-Garcia, J. R.,
Mellouli, S., Nahon, K., Scholl, H. J., 2012.
Understanding smart cities: An integrative framework.
In System Science (HICSS), 2012 45th Hawaii
International Conference on. IEEE. p.2289-2297.
Costamagna, E., Spanò, A., 2013. CityGML for
Architectural Heritage. In Developments in
Multidimensional Spatial Data Models. Springer Berlin
Costamagna, E., Spanò, A., 2012. Semantic models for
Architectural Heritage documentation. In Lecture Notes
in Computer Science. Springer. p.241-250.
Dabove, P., Manzino, A. M., Taglioretti, C., 2014. GNSS
network products for post-processing positioning:
limitations and peculiarities. In Applied Geomatics.
De Laat, R., Van Berlo, L., 2011 Integration of BIM and
GIS: The Development of the CityGML GeoBIM
Extension. In Advances in 3D Geo-Information
Sciences. LNG&C. p.211-225.
Doerr, M., Hiebel, G., Eide, Ø., 2013. CRMgeo: Linking
the CIDOC CRM to geoSPARQL through a
spatiotemporal refinement. In Institute of Computer
Science, Tech. Rep. GR70013.
Doerr, M., Ore, Ch.E., Stead, S., 2007. In The CIDOC
Conceptual Reference Model - A New Standard for
Knowledge Sharing. In Tutorials, posters, panels and
industrial contributions at the 26th International
Conference on Conceptual Modeling. ACS. 83. p.51-
Donadio E. and Spanò A. (2015). Data Collection and
Management for Stratigraphic Analysis of Upstanding
Structures. In Proceedings of the 1st International
Conference on Geographical Information Systems
Theory, Applications and Management. ISBN 978-989-
758-099-4. p.34-39. DOI:
ESRI, 2008. The Multipatch Geometry Type. ESRI White
Paper. Available from:
tch-geometry-type.pdf [Accessed on 12/11/2015.
Fonseca, F., Egenhofer, M., Davis, C., Camara, G., 2002.
Semantic granularity. In Ontology-driven geographic
information systems. AMAI Annals of Mathematics
and Artificial Intelligence, 36 (1-2). p.121-151.
Guizzardi, G., 2005. Ontological foundations for structural
conceptual models. CTIT, Centre for Telematics and
Guarino, N., Oberle, D., Staab, S., 2009. What is an
Ontology?. In Handbook on ontologies, Springer Berlin
Janssen, M., Charalabidis, Y., Zuiderwijk, A., 2012.
Benefits, adoption barriers and myths of open data and
open government. In Information Systems
Management, 29(4). p.258-268.
Kolbe, T. H., Gröger, G., Plümer, L., 2008. CityGML–3D
city models and their potential for emergency response.
In Zlatanova, S., Li, J. (eds.). Geospatial information
technology for emergency response. CRC Press.
Laurini, R., 2015. Geographic Ontologies, Gazetteers and
Multilingualism. In Future Internet. 7(1). p.1-23.
Laurini, R., 2014. A conceptual framework for geographic
knowledge engineering. In Journal of Visual
Languages & Computing. 25(1). p.2-19.
Laurini, R., Thompson, D., 1992. Fundamentals of spatial
information systems. Academic Press.
Mennis J.L., 2003. Derivation and implementation of a
semantic GIS data model informed by principles of
cognition. In Computers, Environment and Urban
Systems. 27. p.455-479.
Métral, C., Falquet, G., Cutting-Decelle, A. F., 2009.
Towards semantically enriched 3D city models: an
ontology-based approach. In Academic track of
Myers D., Avramides Y., Dalgity A., 2013. Changing the
Heritage inventory paradigm, The ARCHES Open
Source System. In Conservation Perspectives, The GCI
OGC, 2012. CityGML UML diagrams as contained in
CityGML Encoding Standard Version 2.0, OGC Doc.
Ronzino, P., Niccolucci, F., Felicetti, A., Doerr, M., 2015.
CRMba a CRM extension for the documentation of
standing buildings. In International Journal on Digital
Schaffers, H., Komninos, N., Pallot, M., Trousse, B.,
Nilsson, M., Oliveira, A., 2011. Smart Cities and the
Future Internet: Towards Cooperation Frameworks for
Open Innovation. In Future Internet Assembly. 6656.
Scholl, M., Voisard, A., 1992. Object-oriented database
systems for geographic applications: an experiment
with O2. In Proc. Int. Workshop on Database
Management Systems for Geographical Applications.
Solovyov, S. A., 2012. Categorical foundations of variety-
based topology and topological systems. In Fuzzy Sets
and Systems. 192. p.176-200.
Van den Brink, L., Janssen, P., Quak, W., & Stoter, J. E.
(2014). Linking spatial data: automated conversion of
geo-information models and GML data to RDF. In
International Journal of Spatial Data Infrastructures
Research. 9, 2014.
Van den Brink, L., Stoter, J. E., Zlatanova, S., 2012.
Modelling an application domain extension of
CityGML in UML. In ISPRS Conference 7th
International Conference on 3D Geoinformation, The
International Archives on the Photogrammetry, Remote
Sensing and Spatial Information Sciences. XXXVIII-4,
part C26, 16–17 May 2012, Québec, Canada. ISPRS.
http://inspire.ec.europa.eu/ [Accessed on 11/11/2015]
www.opengeospatial.org/ [Accessed on 11/11/2015]
[Accessed on 12/11/2015].
www.w3.org [Accessed on 09/11/2015].
http://www.w3.org/TR/rdf-sparql-query/ [Accessed on
www.w3.org/XML/ [Accessed on 09/11/2015].
www.w3.org/RDF/ [Accessed on 10/11/2015].
www.opengeospatial.org/standards/gml [Accessed on
https://it.wikipedia.org/wiki/COLLADA [Accessed on
www.w3.org/2004/OWL/ [Accessed on 11/11/2015].
https://it.wikipedia.org/wiki/COLLADA [Accessed on
http://www.geonames.org/ [Accessed on 12/11/2015].
http://www.icomos.org/en/charters-and-texts [Accessed on
http://vocab.getty.edu/ [Accessed on 12/11/2015].
http://archesproject.org [Accessed on 09/11/2015].
[Accessed on 10/11/2015].
s [Accessed on 09/11/2015].
ge [Accessed on 16/11/2015].
[Accessed on 17/11/2015]
s [Accessed on 17/11/2015]
=700000350 [Accessed on 15/02/2016].
[Accessed on 15/02/2016].