2015 ASCE PublishedPaper
2015 ASCE PublishedPaper
net/publication/280612643
CITATIONS READS
114 4,332
3 authors:
John Haymaker
Perkins & Will
91 PUBLICATIONS 2,228 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Ebrahim P. Karan on 03 August 2015.
Abstract: When making design and construction decisions, planners must consider information from different scales and domains.
Downloaded from ascelibrary.org by Pennsylvania State University on 07/28/15. Copyright ASCE. For personal use only; all rights reserved.
Currently, building and geospatial data are shared and exchanged through a common data format, such as industry foundation classes (IFC).
Because of the diversity and complexity of domain knowledge across building information modeling (BIM) and geographic information
system (GIS) systems, however, these syntactic approaches are not capable of completely sharing semantic information that is unique in each
system. This study uses semantic web technology to ensure semantic interoperability between existing BIM and GIS tools. The proposed
approach is composed of three main steps: ontology construction, semantic integration through interoperable data formats and standards, and
query of heterogeneous information sources. The completeness of the methodology is validated through a case study. DOI: 10.1061/(ASCE)
CP.1943-5487.0000519. © 2015 American Society of Civil Engineers.
Author keywords: Building information modeling (BIM); Geographic information system (GIS); Semantic interoperability; Semantic web
technology.
Introduction The GIS has been used successfully to solve the complexities of
preconstruction planning and to support the wide range of spatial
Building information modeling (BIM) represents building elements analysis used in the logistics perspective of the construction activ-
such as beams, columns, and walls as smart three-dimensional (3D) ities. It enables addressing questions such as: Where are the optimal
objects that include embedded data such as geometry details, locations for the temporary facilities on a construction site? How
energy use data, and lifecycle cost information. BIM provides de- can the construction materials be tracked and monitored through
tailed information for designers and managers and can help answer their supply chain? Where are the dangerous or hazardous areas
questions such as: What is the quantity of each building object or on a jobsite? The integration of BIM and GIS can offer substantial
component? Can a given building design be constructed within benefits to manage the planning process during the design and con-
budget? What is the impact of a given design change on the overall struction phases. While BIM systems focus on developing objects
project scope and schedule? Engineers in the design and construc- with the maximum level of detail in geometry, GIS are applied to
tion community can use BIM to manage design geometry and visu- analyze the objects, which already exist in the physical environ-
alize the model in two-dimensional (2D) and 3D views and together ment, in a most abstract way. The major difficulty in integrating
with the fabricators and owner share and assess various design BIM and GIS systems reflects their incompatibility such as the
options from cost, constructability, and engineering perspectives. modeling environment and reference system (e.g., GIS data are
Sometimes due to the lack of spatial analysis capabilities in BIM, georeferenced and usually two dimensional while the BIM data are
building data are incorporated in the form of an input into a geo- three-dimensional objects located within local coordinate systems).
graphic information system (GIS) tool to support the diversity of Although these two technologies have evolved from distinctly
spatial relationships between topographic and temporary objects. different beginnings, both can benefit from each other if they could
While this integration indicates the presence of a gap in analyzing exchange data effectively. As BIM technology is mainly centered
and processing spatial data within a BIM system, it also indicates on indoor environments, GIS can extend the benefits and appli-
the potential value of an integrated BIM-GIS model that can be cability of existing building models to the outdoor environment.
used to enhance the current practice of data sharing between the However, it is not an easy task to transfer data from BIM to GIS
tools used in the procurement and construction processes. or vice versa without consideration of data format and meaning.
Current state-of-the-art BIM (or GIS) tools enable the data ex-
1
Assistant Professor, Dept. of Applied Engineering, Safety and change between the systems by using a common data format.
Technology (AEST), Millersville Univ., 40 East Frederick St., Millersville, Therefore, the users are able to access data from a different software
PA 17551 (corresponding author). E-mail: Ebrahim.Karan@millersville program and exchange data within the BIM (or GIS) domain. How-
.edu ever, it requires the user to have a thorough understanding of both
2
Associate Professor, School of Building Construction, Georgia systems and their functionalities. The integration tools and current
Institute of Technology, 280 Ferst Dr., 1st Floor, Atlanta, GA 30332. standards lack the ability to help the user to convey meaning, which
E-mail: [email protected] is interpretable by both construction project participants as well as
3
Assistant Professor, Schools of Architecture and Building Construc- BIM and GIS tools. In order to fully integrate GIS and BIM, there is
tion, Georgia Institute of Technology, 280 Ferst Dr., 1st Floor, Atlanta,
a need to provide interoperability at the semantic level.
GA 30332. E-mail: [email protected]
Note. This manuscript was submitted on November 5, 2014; approved
The current approach to exchange and share building data
on June 16, 2015; published online on July 22, 2015. Discussion period between BIM applications is based on the exchange of industry
open until December 22, 2015; separate discussions must be submitted foundation classes (IFC) files. While this approach was, and still
for individual papers. This paper is part of the Journal of Computing remains, an effective way to hold and exchange data among various
in Civil Engineering, © ASCE, ISSN 0887-3801/04015043(11)/$25.00. participants in a building, construction, or facility management
scription framework (RDF) and web ontology language (OWL) understood, and processed by both tools. The ontology construc-
are based on description logics. As description logics describe the tion and semantic integration are selected as the focus of this study.
domain in terms of concepts, roles, and individuals, OWL describes Thus, the literature review is divided into two sections; First, some
that in terms of classes (instead of concepts), properties (instead of of the benefits derived from the integration of BIM and GIS are
roles), and individuals. In particular, the formal specification of the provided. Then, the second section provides further details on the
OWL was influenced by description logics and its RDF/XML ex- ontology-based approaches and semantic web technology.
change syntax was influenced by a requirement for upwards com-
patibility with RDF (Horrocks et al. 2003).
Literature Review
to evaluate construction performance and to facilitate monitoring and retrieval. Based on this project, El-Gohary and El-Diraby
of repetitive construction progress. As one of the last stages, facility (2010), and El-Diraby and Osman (2011) presented an ontology
management is about planning and managing the life cycle of a for the infrastructure and construction domain that relates to con-
building. Obviously, facility managers need a massive amount of struction aspects of infrastructure products.
information for their work. Karan and Irizarry (2014) developed a Despite the successful applications of e-COGNOS identified in
spatial BIM-GIS framework, which can offer facility managers an their research, the next step is to develop a formal ontology for
integrated tool to manage the maintenance and repair processes of construction that allows a user to share and manage domain knowl-
facility management. Liu and Issa (2012) utilized both technologies edge. e-COGNOS is focused on what is called domain ontology,
for detecting and mapping pipe network information. Although the which represents shared concepts in the AEC domain. However,
study mainly relied on BIM and GIS visualization capabilities, it to integrate different types of information, a higher level of details
showed how facility managers can benefit from an implementation is necessary within the ontology. If we define taxonomy as a set
of BIM in geospatial context. of terms and their definitions that are organized by a hierarchy,
ontology provides a framework for representing a concept by its
position in the hierarchy and its relationships to other concepts.
Ontology-Based Approaches for Improving the The result of the e-COGNOS project was a pure taxonomy that
Interoperability only contains construction terms and their relations in a taxonomic
There have been many applications of ontology-based approaches tree. Further research is required to explore the full capability and
in civil and construction engineering. Yurchyshyna and Zarli benefits of mapping this construction-specific taxonomy with other
(2009) presented an ontology-based method for the formaliza- ontologies.
tion and application of construction conformance requirements for Despite the contributions and practical features of these
effective code checking. Wang and Boukamp (2009) adopted onto- ontology-based approaches, there is still no guarantee that hetero-
logical modeling to organize essential concepts of job hazard analy- geneous information sources can be integrated into one system.
sis knowledge and identify applicable safety rules. Elghamrawy Research on the potential application of ontology-driven approaches
et al. (2009) developed a framework that relies on the use of con- on the integration and interpretation of heterogeneous information
cept ontologies for describing and indexing the construction- resources has been recognized only recently in the AEC literature.
problem context information captured through the use of RFID. One of the few relevant researches is the attempt of El-Gohary and
In a similar effort, Sørensen et al. (2010) created a digital link El-Diraby (2011) to develop an ontology merger (Onto-Integrator)
between the virtual models and the physical components in the con- based on semantic similarity comparison methods to merge con-
struction by means of RFID technology and reviewed existing cept taxonomies and ontological relations of source ontologies into
ontologies for information sharing between trading partners, easy an integrated combined ontology. Previous attempts of ontology
access of information, and reading of data stored in electronic tags. development in the AEC have undoubtedly paved the way for
Wang et al. (2011) proposed an ontology-based approach to facili- seamless integration of building and construction related data,
tate the management of context-sensitive construction information however, no application ontology exists for the building and con-
that is stored in different textual documents. Zhong et al. (2012) struction domain that encompasses all IFC classes with different
proposed an ontology for construction quality inspection and evalu- attributes. A gap exists on how to represent BIM and GIS data (and
ation, CQIEOntology, for improving the support to the construction their semantics) in such as a way that can be shared, understood,
quality inspection and management. Park et al. (2013) proposed a and processed by both tools. The objective of this paper is to bridge
conceptual system framework for the proactive defect management this gap through the development of an application ontology that
with three interrelated system solutions, namely defect data col- contains a set of building and geospatial terms (e.g., building ele-
lection template, defect domain ontology, and augmented reality. ments, topography, and geolocations) and their semantics.
Another study used BIM data and ontology to automate the selec-
tion and matching work items to the elements of buildings and their
materials (Lee et al. 2014). Karshenas and Niknam (2013) devel- Research Methodology
oped an ontology-based approach to facilitate project information
sharing between design and estimating domains. The methodology for extending BIM interoperability to the geospa-
Anumba et al. (2008) reviewed examples and case studies of tial domain consists of five stages. First, an IFC-compliant ontology
ontology-based information and knowledge management systems describing the hierarchy structure of BIM objects, their relation-
in the construction delivery process and found that middleware ap- ships, and their properties is developed. The emphasis is on seman-
plications, such as semantic web, have the potential to meet some of tic indexing and retrieval of building information from an IFC
the technical challenges inherent in the development and use of model. This study makes use of existing GIS ontologies that ob-
ontologies for construction information. The e-COGNOS Project viate the need for the transformation of GIS schema into ontology.
represent the natural classes of the IFC domain where only neces-
sary conditions are specified and they can be recognized by their
definition. Second, defined concepts are used to represent sub-
classes of the primitive ones (i.e., built using primitive concepts and
Second, ontology mapping is used to link similar relationships properties). Thus, the authors define the IFC classes at the top of
or concepts between the source (e.g., BIM) and target (e.g., GIS) the hierarchical structure of the ontology as primitive concepts.
ontologies. The output is an extended ontology that contains all For example, assume that an individual x is an instance of Ifc-
classes and properties from both BIM and GIS domains that are Window (as a primitive concept), thus x possesses the properties of
relevant to the case study. Then, building’s elements and GIS data IfcWindow such as overall height and overall width. The standard
are translated into semantic web standards, and thus can be proc- window, which is inserted into an opening and its profile repre-
essed by semantic web applications. Once the information has been sents a rectangle within the 2D plane of the opening, is defined
gathered from different sources and transformed into an appropriate by IfcWindowStandardCase. The authors define this IFC entity
semantic web format, the SPARQL query language is used in the as a defined concept, so the associated properties of the IfcWindow-
fourth section to retrieve this information from a dataset. Finally, StandardCase are necessary and sufficient. Again, assume that an
the completeness of the methodology is validated through a case individual y is an instance of IfcWindowStandardCase (as a defined
study. Fig. 1 demonstrates the process of establishing semantic concept), thus y possesses the properties of IfcWindowStandard-
interoperability and integration and the activities supported by each Case and the y individual that possesses the set of the associated
stage. Each step in the process is explained in detail in the following properties of IfcWindowStandardCase (e.g., inserted into an open-
sections. This paper adopts ontology mapping methods that are ing, etc.) suffices to be inferred as an instance of the IfcWindow-
being increasingly used to map BIM and GIS ontologies in the StandardCase.
second step, Integration, and the process developed by Karan For the purpose of forming the ontology, the authors define the
et al. (2015) to translate semantic web query results into the XML IFC classes (or concepts in description logic) by their supertype
representations of the IFC schema and data in the fourth step, entities and their relations with the other classes. Continuing with
Manipulation. the above examples, IfcWindow can be defined as shown in Fig. 2.
It defines a primitive concept IfcWindow, which is a subtype
of IfcBuildingElement. The keyword defprimclass is used to
Step 1: Conceptualization define the primitive concepts and introduce a set of necessary
but not sufficient conditions. This expression also states that all
This section describes how the IFC schema can be elevated to an
IfcWindow classes have at least one OverallHeight that is a positive
ontological level by using description logic. Depending on the
measure, greater than zero. The defined concept IfcWindow-
components and the level of detail, ontologies can be divided into
StandardCase is defined as [defconcept IfcWindowStandardCase
generic (or upper), domain, and application ontologies. A generic
(?w IfcWindow) : : : ], where defconcept creates named descriptions
ontology describes general concepts such as space, time, role,
that describe sets or classes of objects.
object, action, etc., which are applicable across a wide range of
The authors use OWL ontologies to create the application on-
domains. Domain ontology is created with the aim of formalizing
tology (hereafter referred to as the BIM ontology). The OWL
and representing shared concepts in a specific domain of interest
axioms provide semantics about classes and properties by assigning
(e.g., AEC). For instance, a rule about a specific role can be rep-
necessary and/or sufficient characteristics to a class. The Sub-
resented: A construction worker uses a tape measure to take a meas-
ClassOf axiom represents subclass/superclass relationship, so since
urement; where construction worker is an instance of the concept
IfcWindow is a subclass of IfcBuildingElement, it necessarily in-
worker, tape measure is an instance of the concept measuring tool,
herits all characteristics of IfcBuildingElement, but not the other
and uses and take a measurement are used to identify relations
way around. The primitive concepts introduced by defprimconcept
between these conceptual elements. The e-COGNOS (as well as are translated to OWL with subclassOf axioms. The Equivalent-
many other examples mentioned earlier) is a good example of Classes axiom states that two or more class expressions consist
such ontologies. Application ontology is a representation of the se- of the same set of individuals, so they are equivalent to each other
mantics of a specific, focused application domain, which defines
relevant concepts for a particular application (e.g., BIM or GIS).
Given the increasing role of application ontologies in facilitating (defprimclass IfcWindow (?be IfcBuildingElement)
the integration of different types of information, this step examines :=> (and (exists (?oh)
how this level of ontology can be used to provide semantic inter- (and (OverallHeight ?oh)
operability between BIM and GIS operations. The focus is on IFC (>= (OverallHeight ?oh) 0)))
schema items such as attributes, classes, data types, individuals,
Fig. 2. IfcWindow defined in description logic
and relations.
Step 2: Integration
Ontology mapping is used in this step to determine semantically
corresponding entities among BIM and GIS ontologies. Using an
example, the authors provide a brief description of the process. For isDefinedBy
rdf:resource
further details refer to Balachandar et al. (2013) and Hu et al. #geopoint
(2005). Assume that individuals (as BIM users) want to integrate (b)
topographic data that are defined in GIS ontology from the Center
of Excellence for Geospatial Information Science (CEGIS) (USGS Fig. 4. (a) RDF graph (reprinted from Automation in Construction,
2013) with their BIM ontology. In order to measure the structural Vol. 53, Ebrahim P. Karan and Javier Irizarry, “Extending BIM inter-
similarities between BIM and GIS ontologies, the authors adopt operability to preconstruction operations using geospatial analyses
Graph Matching for Ontologies (GMO) approach. This ontology and semantic web services,” pp. 1–12, Copyright 2015, with permis-
matching approach uses RDF bipartite graph model, which was sion from Elsevier); (b) bipartite graph of the GIS ontology example
first introduced by Hayes and Gutierrez (2004), to represent ontol- (instance labels have been omitted for clarity)
ogies. Fig. 3 shows the RDF graph and bipartite graph of the GIS
ontology. In the bipartite graph, property nodes are represented by
circles, class statements are represented by a rounded rectangle, and
edge labels S, P, and O indicate their subject, predicate, and object. subClassOf
The adjacency matrix of the directed bipartite graph of ontology, isDefinedBy
denoted by A, has the following block structure: hasAttribute
0 1 Vegetation 1 0 0 0
0 0 AES Surface Water 0 1 0 0
B C
A¼@ 0 0 AS A Topography 0 0 0 0
Terrain 0 0 1 0
AE AOP 0 #Geo:Point 0 0 0 1
N1 1 0 0 0 0 1 0 0
where AES = matrix representing the connections from external en-
N2 1 0 0 0 0 1 0 0
tities (e.g., rdfs:subClassOf) to statements; AS = matrix represent- N3 1 0 0 0 0 1 0 0
ing the connections from ontology entities (internal entities) N4 0 1 0 0 0 0 0 1
to statements; AE = matrix representing the connections from state-
ments to external entities of the ontology; and AOP = matrix rep- Fig. 5. Matrix representation of GIS ontology
resenting the connections from statements to internal entities.
In the example in Fig. 3, the external entities include some
common ones (e.g., rdfs:subClassOf, rdfs: isDefinedBy) used in
two ontologies. However, If those entities are not used as subjects Now, it is possible to represent the similarity matrix of BIM
in the Ontology (as in the current example), AES is a zero matrix. ontology entities to GIS ontology entities and the external entities
The matrix representation of GIS ontology in Fig. 4 is shown of GIS ontology to the external entities of BIM ontology. Based on
in Fig. 5. the formulation in Hu et al. (2005), the structural similarity matrix
as shown in Fig. 9.
and then transferred into GIS software as a metadata format for There are some IFC entities such as IfcNormalisedRatio-
building and geospatial data description. Unlike BIM, the coordi- Measure, IfcRatioMeasure, and IfcSpecularExponent, that can be
nate system is georeferenced in a GIS environment. In order to defined with single attribute and without any relation to another
address the issue of positioning the building within GIS context, IFC entity. Therefore, these distinct entities are not defined as
spatial coordinates is defined and transformed from Cartesian to an OWL class, instead they are defined an OWL data-type property.
Geospatial (i.e., georeferenced) with the aid of a coordinate trans- Similar to the role of rdfs: subClassOf in forming the taxonomy
formation matrix at the beginning of the process. The header of an
of IFC classes, taxonomy of properties is formed by rdfs:sub-
IFC file states its name, description, translator version (if used), and
PropertyOf. Since the distinct IFC entities are used to specify an
schema version. The IFC entities and their attributes (both normal
IFC attribute, the authors define them as subPropertyOf other IFC
and optional) are specified in the body of the IFC file. Each IFC
attributes. For instance, IfcMeasureWithUnit entity has two simple
entity starts with a number sign (#) character followed by a number.
type attributes: (1) ValueComponent defined by IfcRatioMeasure
Since every individual in OWL needs a unique identifier, the au-
distinct entity, and (2) UnitComponent defined by IfcSIUnit entity.
thors will use these unique numbers to define the common instan-
ces or individuals of different classes. Each new IFC entity is The ValueComponent attribute of the IfcMeasureWithUnit is de-
defined after an equal sign, (=), and its attributes are represented fined as shown in Fig. 10.
by a set of comma separated values within parentheses. The normal One or more subattributes are needed to describe the properties
attributes for the IFC entity always get nonnull values, while the and values of a complex type attribute. For instance, IfcPropertySet
optional attributes may have null values indicated by a dollar sign. has one complex type attribute, HasProperties, which is defined
Regardless of the type of attributes, each IFC entity is represented with a set of 1 to some subattributes (i.e., IfcPropertySingleValue).
as the subject of the RDF statement using an owl:Class and use a These subattributes are defined either by distinct IFC entities or
rdf:about statement to describe that subject. Also, the rdfs:sub- by a set of individuals. Again, subPropertyOf is used to define the
ClassOf property is used to state that one IFC entity is a subtype distinct IFC classes. The IfcPropertySingleValue attribute of the
of another entity or resource. Consequently, every OWL class is a IfcPropertySet is defined as shown in Fig. 11.
subclass of owl:Thing. Fig. 12 shows the flowchart of IFC to RDF/OWL translation
An IFC attribute can be defined by a value parameter or an IFC process for different types of IFC attributes. This process continues
entity. In order to reflect the attribute type, the IFC attributes are until no more attribute exists for further translation as shown
divided into three groups in the conversion process: (1) leaf node, in Fig. 12.
(2) simple type, and (3) complex type. A leaf node attribute is de-
fined by a value parameter. For example, the IfcOrganization entity
has five leaf node attributes; Id (optional), Name (normal), Descrip- <owl:NamedIndividual rdf:about="…Name">
tion (optional), Roles (optional), and Addresses (optional). Thus, <…Name rdf:datatype="&xsd;string">Autodesk Revit 2014 (ENU)</…Name>
</owl:NamedIndividual>
this IFC entity can be defined as #1= IFCORGANIZATION
($, ‘Autodesk Revit 2014 (ENU)’, $,$,$). The IfcOrganization Fig. 8. String value of the Name attribute defined as OWL
entity is defined as shown in Fig. 6.
The type of literal data is defined by rdfs:range and rdfs:domain
is used to state that the leaf node attribute is an instance of the <owl:DatatypeProperty rdf:about="…ApplicationDeveloper">
IFC entity. Thus, the authors define the Name attribute of the <rdfs:domain rdf:resource="…IfcApplication"/>
IfcOrganization entity as shown in Fig. 7. <rdfs:isDefinedBy rdf:resource="…IfcOrganization"/>
</owl:DatatypeProperty>
<owl:Class rdf:about="..IfcOrganization">
Fig. 9. ApplicationDeveloper attribute of the IfcApplication defined
<rdfs:subClassOf rdf:resource="…IfcEntity"/>
</owl:Class> as OWL
property? value {?subject owl:Datatype Property OverallHeight.? component data are exchanged frequently between these two tech-
subject? property? value.} would return the value components of nologies. In addition to this high level of integration, using the same
OverallHeight attribute in the dataset. The authors adopt the frame- case study as the baseline for comparison is another reason for
work developed by Karan et al. (2015) to represent semantic web choosing the CSCM application as the validation approach to
query results as ifcXML building models. Thus, the resulting evaluate the benefits of the proposed methodology. The methodol-
ifcXML document can be loaded into a BIM authoring tool. Also, ogy is employed for monitoring CSCM of a building project in
CSV format is used for expressing the results of a select query Carrollton, Georgia, namely, The School of Nursing at the Univer-
in GIS. sity of West Georgia. The project involved a three-story, 65,000-ft2
…<owl:Class rdf:about="…IfcPropertySet">
<rdfs:subClassOf rdf:resource="…IfcPropertySetDefinition"/> building components included in the BIM model are combined
</owl:Class>
… Complex Type
with network analysis in a GIS. The authors use either IfcElement-
…<owl:DatatypeProperty rdf:about="…HasProperties"> Attribute Quantity or IfcPropertySet to define a set of quantities (e.g., length,
<rdfs:domain rdf:resource="…IfcPropertySet"/> (CType: Set) height, gross footprint area, etc.) of an element’s physical property.
<hasSetOf rdf:resource="…IfcPropertySingleValue"/>
</owl:DatatypeProperty> Users of the proposed methodology should be aware of hierarchies
of attributes and alternatives to define the values and quantities.
For instance, it is possible to retrieve area, volume, and length
Fig. 13. IFC entities transformation into RDF/OWL classes in the quantities for IfcCurtainWall from its IfcPropertySet, and Height,
case study Length, Width, GrossFootPrintArea, and GrossVolume for IfcWall-
StandardCase from its IfcElementQuantity. Up to this point, the
supplier locations and material properties defined in the BIM model
are combined together in the GIS model.
building accommodating all functions for nursing education and
The material status is created as instance parameters in the BIM
support spaces.
model and assigned to all categories like walls, windows, doors,
The authors define the relevant parameters for each type of
and columns. All these parameters are defined as a Date–Time var-
building materials (e.g., engineered-to-order (ETO), made-to-order
iable and have two entries: one for schedule and one for actual date.
(MTO), etc.). These relevant parameters (or new attributes) will
Here, the authors take advantage of the manipulation step of the
be used to determine the status of each material. Therefore, Ifc- research methodology to translate semantic web queries to seman-
PropertySingleValue is used to add cost and schedule data and tically equivalent IFC entities. The schedule and actual date param-
supplier information into the IFC in addition to the quantities eters are leaf-node attributes, however, their corresponding IFC
(e.g., area, volume, weight) specific to each element type. The class might have simple or complex type attributes. Then, the query
IfcPropertySingleValue has four attributes: (1) Name is used to de- results are converted into ifcXML that can be imported into a BIM
fine the name of the new parameter, (2) Description can be used to model. Following this, the user can compare the actual dates with
provide more information about the parameter, (3) NominalValue is the schedule dates to categorize elements based on the delays in
used to assign a single value, and (4) Unit can be used to further delivery. The RDF/OWL documents generated by the proposed
describe the NominalValue. Once defined, the IfcPropertySingle- methodology were validated using the W3C validation service.
Value is used as a subattribute for IfcPropertySet in order to define It addition to the syntax and structure, the authors also checked the
all extensible properties that apply to specific building products. semantics of the generated models in the BIM environment.
Fig. 13 shows the IFC entities and their corresponding RDF/OWL Current practice and the proposed methodology were used in
classes translated by the proposed method. parallel during this case study. Challenges in BIM and GIS data
The description of curtain wall is used as an example of the sharing and interoperability using existing approaches are twofold:
ETO product. Curtain wall is represented as an IfcCurtainWall 1. Data models related to building or geospatial components are
element and implemented as a subclass of IfcBuildingElement. specified and structured in different formats, thus, different
The IfcCurtainWallType is used to define the specifications of BIM and GIS authoring tools cannot exchange and share their
curtain wall, such as the shared properties that are common to all data models between each other; and
curtain wall types, the optional properties that are common to cer- 2. Even if some BIM-related information are transferred to GIS
tain types of curtain wall, and new properties that are added to or vice versa (e.g., using a data conversion tool), there is no
manage the supply chain. A typical IfcCurtainWallType entity is guarantee that another system can interpret the data being
represented as follows (In the description below, texts in brackets transferred.
[] refers to the attribute names and they are not presented in the IFC Based on the results of the case study, it can be concluded that
document): the semantic web technology enables BIM users to represent
#[line number goes here]=IFCCURTAINWALLTYPE (Step 1), share (Steps 2 and 3), and discover (Step 4) building and
([GlobalId as a normal and leaf node attribute], [OwnerHistory GIS data through ontologies. Table 2 provides a comparison be-
as a normal and simple type attribute], [Name as an optional and tween the proposed approach and state-of-the-art BIM and GIS
leaf node attribute], [Description as an optional and leaf node attrib- tools, including Autodesk’s AutoCAD and Revit, Bentley’s Archi-
ute], [ApplicableOccurrence as an optional and leaf node attribute], tecture, Graphisoft’s ArchiCAD, and Esri’s ArcGIS, based on the
([HasPropertySets as a normal and complex type attribute]), fraction of building and GIS features that can be exchanged (or
[RepresentationMaps as an optional and complex type attribute], supported) between BIM and GIS models without losing their
[Tag as an optional and leaf node attribute], [ElementType as an semantics (this table only shows some examples of these features).
optional and leaf node attribute], [PredefinedType as a normal These features are limited to those used for the case study, and
and leaf node attribute]); divided into building elements, geometry elements and basic
constructs. In order to understand what fraction of the semantics be partially transferred between BIM and GIS tools. In contrast, the
are understood (or returned) by the destination system, the recall proposed approach partially recalled around 42% (26 out of 62) of
index is categorized as full, partial, and none. If the semantics can the BIM and GIS semantics. A large portion of the semantics that
be retrieved in the case of two-way exchanges (i.e., from BIM to cannot be recalled comes from the data items that are not related to
GIS and back again), the BIM and GIS features are fully recalled. GIS applications (e.g., chiller, column, curtain wall). Moreover, the
However, if the semantics of the features can be delivered only full-recall rate using the proposed approach is considerably higher
in one-way exchange, they are partially recalled. For example, than that for the existing tools. According to the results of the
in order to transfer a geometric surface from a building model to validation step, around 40% of the semantics are retrieved using
a GIS model (or vice versa), the user needs to represent the feature the proposed approach in the case of two-way exchanges, while
as 2D CAD data and then annotate it with appropriate keywords. only around 10% can be conveyed using the state-of-the-art tools.
However, this practice does not convey the meaning of the geom- These results provide evidence for the effectiveness of the proposed
etry element in the case of two-way exchanges, and the BIM user approach for extending the interoperability between the building
will need to define a new element after it transfers back to the build- modeling and geospatial analysis tools.
ing model. Although this feature is partially recalled by state-
of-the-art tools, the proposed approach fully recalls the geometric
surface feature. There is no recall when the semantics cannot be Limitations of System
shared and reused across BIM and GIS applications. Almost two
third of the features used in this study could not be semantically The current lack of formal AEC ontologies can lead to incon-
shared by state-of-the-art tools, and only 24% (15 out of 62) can sistencies between the ontologies developed by multidisciplinary
Slab X X
Window — — X — X — manually imported into the GIS model. In contrast, the proposed
Geometry elements method enables the user to query the content of the data sources.
Bounding edges — — X — X — Further work needs to be done to develop globally-agreed ontol-
Elevation — — X X — — ogies for the construction domain.
Geometric surface — X — X — —
While the results are encouraging for the case study, additional
Point X — — X — —
Rectangular X — — X — — case studies would be necessary to examine the applicability of
coordinate system semantic web technology to the AEC domain. The fraction of
Volume — — X X — — building and GIS features that can be exchanged by the proposed
Basic constructs approach in comparison with the state-of-the-art tools (Table 2) rec-
Asset — — X — — X ognizes the semantic web as a key-enabler for integration of data in
Calendar date — — X — X the construction process. It is expected that the proposed approach
Capacity — — X — — X would enable process integration that can lead to improvements
Cost — X — — X — in the exchange of BIM and GIS information. How semantic web
Dimensions of the — — X X — —
technology enables both data and process integration is not ad-
base quantities
Globally unique — X — X — — dressed in the case study. Therefore, it is recommended that the
identifier proposed approach is applied for different use cases that focus
Text X — — X — — on the interaction between the BIM user and the system (process
Units of the base — — X X — — integration). The present methodology demonstrates how to gener-
quantities ate RDF triples from IFC contents or IFC entities from query re-
sults; however, the entire conversion process is not fully automated
and selecting of equivalent (or nearest equivalent) IFC and RDF/
OWL entities is undertaken manually. For instance, once the user
professionals involved in AEC projects. Thus, two ontologies specifies an equivalent IFC entity for a query result manually, then
might have different levels of granularity. When there is no equiv- the proposed manipulation structure translates and fills all query
alent of a concept in the destination system, the proposed approach results into an ifcXML document automatically. A reasonable
cannot fully capture the semantics of the concept. This is why only extension of the present research is to develop a fully-automated
parts of the semantics are recalled. A potential solution is to con- converter in one of the server-side programming languages such as
sider a most similar entity class across the destination ontology; php scripting language.
however, this approach is not applied in the present paper.
Because of the extra information about all of the entries em-
bedded into RDF and ifcXML files, converting building or GIS References
data to RDF or query results into ifcXML can make the file size
very large. The model used in the case study consisted of around Anumba, C. J., Issa, R. R., Pan, J., and Mutis, I. (2008). “Ontology-based
60 features, yet the RDF file was almost 660,000 lines of code with information and knowledge management in construction.” Constr.
a total file size of 36.4 MB. Thus processing all data can be highly Innovation Inf. Process Manage., 8(3), 218–239.
computationally intensive and time-consuming through a web Balachandar, K., Thirumagal, E., Aishwarya, D., and Rajkumar, R. (2013).
portal. Ontologies are also evolving and new concepts and fea- “Ontology mapping techniques and approaches.” Int. J. Comput. Appl.,
tures are added to the AEC and geospatial domains of knowledge. 65(24), 13–20.
Because the proposed methodology adopted ontology mapping Elbeltagi, E., and Dawood, M. (2011). “Integrated visualized time control
techniques, it has the ability to overcome the short-term deploy- system for repetitive construction projects.” Autom. Constr., 20(7),
940–953.
ment obstacles. However, the long-term effectiveness is dependent
El-Diraby, T., Lima, C., and Feis, B. (2005). “Domain taxonomy for
upon developing a data framework that automatically integrates
construction concepts: toward a formal ontology for construction
itself with the globally-agreed ontologies. knowledge.” J. Comput. Civ. Eng., 10.1061/(ASCE)0887-3801(2005)
19:4(394), 394–406.
El-Diraby, T., and Osman, H. (2011). “A domain ontology for construction
Conclusion and Future Work concepts in urban infrastructure products.” Autom. Constr., 20(8),
1120–1132.
The primary objective of this study was to extend the semantic Elghamrawy, T., Boukamp, F., and Kim, H.-S. (2009). “Ontology-based,
interoperability between BIM and GIS tools. The contributions semi-automatic framework for storing and retrieving on-site construc-
of the proposed model to the existing body of knowledge are two- tion problem information—An RFID-based case study.” Construction
fold. The first is to enhance data exchange and integration between Research Congress, ASCE, Reston, VA, 457–466.