|UFDC Home||myUFDC Home | Help|
1 IFC BASED CONSTRUCTION INDUSTRY ONTOLOGY FOR INFORMATION RETRIEVAL FROM BUILDING INFORMATION MODELS By LE ZHANG A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIR EMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2012
2 2012 Le Zhang
3 To my w ife and my p arents
4 ACKNOWLEDGMENTS I would like to thank my committee chair, Dr. Randy Chow, for his continuous and rigorous guidance along the u nusual way of finishing a degree in a field that I was not really familiar with. I also ow e my gratitude to the members of my committee. Dr. Raymond Issa, my external advisor, gives me invaluable support. Dr. Shigang Chen is very kind to sit in my committe e and provides me many scholarly advices on the technical issues. I am also obliged to Mr. John Bowers and Ms. Joan Crisman, whose mastery of the academic system and cheerful motivation make this journey much easier and enjoy able.
5 TABLE OF CONTENTS p age ACKNOWLEDGMENTS ................................ ................................ ................................ 4 LIST OF TABLES ................................ ................................ ................................ ........... 6 LIST OF FIGURES ................................ ................................ ................................ ........ 7 ABSTRACT ................................ ................................ ................................ .................... 8 CHAPTER 1 INTRODUCTION AND RESEARCH OBJECTIVES ................................ ............... 10 2 LITERATURE REVIEW AND RELATED WORKS ................................ ................. 13 Industry Foundation Classes (IFC) ................................ ................................ ........ 13 Building Information Modeling ( BIM ) ................................ ................................ 13 IFC Specification ................................ ................................ ............................. 13 IFC P artial M odel E xtraction ................................ ................................ ............ 15 Ontology ................................ ................................ ................................ ................ 16 Origin and Definition of Ontology ................................ ................................ ..... 16 Ontology Structure ................................ ................................ .......................... 18 Ontology Languages and Web Ontology Language (OWL) ............................. 19 Current Domain Ontology Research in Construction Industry ......................... 22 3 DEVELOPMENT OF IFC BASED CONSTRUCTION INDUSTRY ONTOLOGY .... 27 Scope Definition ................................ ................................ ................................ .... 27 A Closer Look on IFC Specification Structure ................................ ........................ 28 Basi c Ontology Development ................................ ................................ ................ 29 Extended Ontology Development ................................ ................................ .......... 35 4 ONTOLOGY BASED INFORMATION RETRIEVAL FROM BUILDING INFORMATION MODE L S ................................ ................................ ..................... 37 A Simple Ontology based Information Retrieval Use Case ................................ .... 37 The Partial Model Extraction Algorithm ................................ ................................ .. 38 The Prototype Application ................................ ................................ ...................... 44 5 CONCLUSION AND RECOMMENDATIONS ................................ ........................ 47 LIST OF REFERENCES ................................ ................................ .............................. 49 BIOGRAPHICAL SKETCH ................................ ................................ ........................... 54
6 LIST OF TABLES Table page 3 1 Inheritance relation for IfcWi ndow ................................ ................................ ......... 31
7 LIST OF FIGURES Figure page 3 1 IFC2x4 Architecture Diagram ................................ ................................ ............ 30 3 2 The cl ass hierarchy in the ontology ................................ ................................ ... 32 3 3 The properties of IfcWindow ................................ ................................ .............. 32 3 4 The relations of IfcWindow ................................ ................................ ................ 33 3 5 A sample IfcWindow ontology in UML ................................ ............................... 34 3 6 A sample IfcWindow ontology in RDF/XML ................................ ....................... 34 4 1 Tree structure of IFC elements ................................ ................................ .......... 39 4 2 Tree structure of general IFC viewers ................................ ................................ 40 4 3 Ontology enhanced tree struct ure of IFC elements ................................ ........... 41 4 4 Partial model extraction sample ................................ ................................ ......... 42 4 5 System architecture of the application ................................ ............................... 44 4 6 Screenshot of the prototype Java application ................................ .................... 46
8 Abstract o f Thesis Presented t o t he Graduate School o f t he University o f Florida in P artial F ulfillment of the R equirements for the D egree of M aster of S cience IFC BASED CONSTRUCTION INDUSTRY ONTOLOGY FOR INFORMATION RETRIEVAL FROM BUILDING INFORMATION MODELS By Le Zhang May 2012 Chair: Randy Y. C. Chow Major: Computer Engineering The construction indu stry is information intensive, and the application of Information Technology (IT) has brought productivity improvement in each sub field of the construction industry. Building Information Modeling (BIM) is recognized as the new paradigm since Computer Aided Design (CAD). The research and application of BIM has been focused on the entire project and the complete life cycle. However, the daily routine on a construction jobsite has specific requirements and creates certain limitations regarding the usage of inf ormation stored in a BIM model. Such limitations include the scarcity of computing resources and trained personnel. One of the requirements is to view a partial model instead of the original, complete model. To extract a partial model from the original com plete model has been identified as one of the BIM application requirements that need further research. In this research an ontology based method to extract a partial model from a complete BIM model is explored. The partial model, as well as the complete m odel, should be defined in the Industry Foundation Classes (IFC) format, which is the widely supported openBIM standard. An ontology is developed based on the existing IFC
9 schema specifications. For each specific IFC model, an ontology ABox is created comb ining the ontology TBox and the instances in the model. The ABox works as an index for the partial model extraction. A prototype application in Java is developed to demonstrate and validate the extraction process.
10 CHAPTER 1 INTRODUCTION AND RESEARCH O BJECTIVES The construction industry is information intensive (Brandon and Betts 1995; Wang et al. 2010). Information Technology (IT) has been applied in almost all specific areas in the construction industry, resulting in huge improve ment of productivity i n each specific area. Computer Aided Design (CAD) maybe the most popular application. With the advent of Building Information Modeling (BIM), the media for the information storage and exchange in a construction project is evolving from traditional 2D based drawings and documents into electronic models. The transition brings great benefit to the industry, including 3D visualization and direct use of the model in a variety of computer based analyses. However, associated with this trend there are also some pr oblems. The key concept of BIM is a complete object oriented digital model of the building. Besides the geometry information, other building information that is not available in traditional CAD solutions is also stored in the model, including but not limi ted to building materials and costs, project specification and contract information, building components manufacturer, price and warranty, etc. The stored information could be retrieved and reused easily once they are entered into the model, thus eliminati ng the need to re input or even re collect information that is valuable in later stages of the building life cycle. Moreover, the stored information can readily be used for various analysis of the building, such as structural stability, energy consumption or building code compliance (Eastman et al. 2008). However, ambitious and promising as it is, the majority of current BIM applications are used during the building design and pre construction stages. BIM application on construction jobsites, especially on small projects, is rare. One of the reasons is that
11 BIM models are usually computing intensive, while the required computing power and trained personnel may not be available on construction jobsites. In addition, most of the daily work on a construction si te is being accomplished by specialty subcontractors, who only deal with small portions of the project and do not need to access the whole model. There are also other situations that more intelligent and information rich BIM models are purposefully avoide d. For example, in many construction education simulation research projects, the authors resort to gaming engines to generate a 3D environment instead of using dedicated or even already existing BIM models. One of the reasons is that the simulation may onl y need the exterior dimension and texture information of a building or a room, but the extra information stored in a BIM model makes it cumbersome and hard to manipulate. Currently most of the research and application of BIM have been focused on adding in formation to the model so that it could be used on the entire project and through out its complete life cycle. Actual application scenarios of BIM models, such as daily construction routine on a construction jobsite or construction education simulation, etc ., may have specific requirements regarding the usage of certain partial information stored in a BIM model. In a recent review by Tizani and Mawdesley (2011), it is Hen ce, how to filter redundant information for specific scenarios and extract designated information from a complete BIM model is another important issue to make full use of the BIM models.
12 This study proposes an ontology built with Web Ontology Language (OW L) based on Industry Foundation Classes (IFC) specifications to help the information retrieval process from an IFC model. With the semantic advantage provided by the ontology, the information retrieval process could directly query the IFC model without inv olving the complications of IFC specification. The development of the ontology and the partial model extraction algorithm based on the ontology is discussed, and a prototype Java application is developed to demonstrate the process.
13 CHAPTER 2 LITERATURE R EVIEW AND RELATED WORKS Industry Foundation Classes (IFC) Building Information Modeling ( BIM ) Different from the traditional CAD applications, the building blocks of a BIM model as well as its drawings are not 2D lines and shapes, but object oriented 3D e lements representing the actual building blocks of an actual building. The elements are defined by parameters that specify the different properties of an element. A wall is defined only once, with its length, width, and height as the object properties (als o called parameters). Different views of the wall are all generated from this central representation, hence eliminating redundancies. A BIM model is such a digital parametric model of the building that aggregates all the object oriented building elements. A BIM model is expected to be reused in the whole life cycle of a building, from inception through design and construction, then facility management and finally demolition and disposal. IFC Specification The Industry Foundation Classes (IFC) is the commonl y accepted data exchange standard for BIM. IFC is proposed to address the lack of interoperability in construction IFC is a set of definitions describing the consistent data representation of building components (Liebich 2010). Developed and maintained by buildingSMART (formerly known as International Alliance for Interoperability, IAI), it is designed to be able to store and exchange all building information over the who le building lifecycle. The IFC object specification includes not only the geometric information of the building elements but
14 also their properties and relationships, and endows the IFC objects with intelligence (Vanlande et al. 2008). As an instance of th e International Standard Organization (ISO) 10303 international standard, one of the advantages of IFC is that it is an open standard and everyone has full access to the information within. Therefore it is ideal for transferring data between different soft ware platforms. The native IFC format, the EXPRESS modeling language, is based on plain text, and will become quite large if used to store all the building information in one file (Campbell 2007). This is one of the contributing factors that IFC is not bei ng used in any BIM authoring software for internal data storage, computation and real time rendering. The EXPRESS is not easy to work with and ha s only limited research population and supporting tools. EXPRESS G is a more human readable graphical expressio n of the EXPRESS schema. IFC also supports XML format storage. It allows any IFC model to be described in XML format under the ifcXML schema, which is equivalent to the EXPRESS schema, although not 100% of the EXPRESS schema information could be expressed in ifcXML (Liebich 2010). A t the time of writing, the newest version is 2x4 RC 2, available since September 2010. The new version features a new documentation format under ISO documentation requirements, besides other improvements. This new version is goin g to be the draft for a new full ISO standard numbered ISO 16739 (buildingSmart 2010b). The new version of IFC, however, does not yet have an XML schema available. The most mature version of IFC is 2x3 TC1, which has more BIM application s and programming t ools support. The open standard nature of IFC makes the information available to everybody, but the IFC specification itself is too complicated for direct use for even experienced
15 developers without special training. Currently the most widely used method to access information of an IFC model is to use third party application programming interfaces (APIs) such as IFCEngine.dll or OpenIFCTools, which works as an interface between the program and the native IFC file (TNO 2010; Beetz et al. 2010; OpenIFCTools 2011). IFC P artial M odel E xtraction IFC provides a good core reference for a complete, general purpose BIM model, but does not provide specifications on partial model operations, including partial model extraction and merge. The requirement and advantages for partial model extraction has been identified by several researchers. The primary purpose of generating a partial model is to reduce the size and complexity of an IFC model, either to fit into special domain application requirement or data transmission requirement (Beetz et al. 2009; Weise et al. 2003). Two types of partial model could be identified based on different scenarios. The first one is to extract information on a certain view or aspect of the building out of the complete model, e.g. a model wi th only the geometries, or a model with only the information related to green building rating. The Model View Definition (MVD) initiative of buildingSMART is an example of works on such extraction (buildingSMART 2011). The Partial Model Query Language (PMQ L) is an implementation of this kind of extraction in XML format. Adachi (2002) Beetz et al. (2009) proposed a graph query method on the IFC ontology that compartmentalizes the ontology, which could filter out the geometry and topology information, leaving the rest for logic based reasoning. The second kind of partial model is a subset of all the entities in the original model. The Generalized Model Subset Definition (GMSD) schema proposed by Weise et al. (2003) is a good example of such extraction. The wo rk is based on EXPRESS, the
16 native modeling language of IFC. Although the compatibility with IFC is ensured, EXPRESS has a far less research population and readily available tools. Along with the partial model extraction, merging a potentially updated part ial model back into the original model is also addressed in this research. As different as they are, one of the major similarities of the above mentioned methodologies is that all the instances of the original model are kept unchanged. However, in real ap plication scenarios, due to differen t modeling practices, an element might need to be modified to suit the partial model extraction process. One such example is described in the extraction algorithms section of this paper. Ontology Origin and Definition of Ontology The word Ontology (with O capitalized) originated from philosophy. Being a branch of metaphysics that deals with the nature and structure (relation) of reality, it is defined the world (Guarino et al. 2009 ; Studer et al. 1998). Under this definition, the subject of Ontology could even be those without actual existence. The word ontology (with o in lowercase) is regaining its popularity in modern information technology since the be ginning of 1990s, coined by Tom Gruber (Studer 1998). It originated from research of Artificial Intelligence (AI) in knowledge management and knowledge engineering community, focusing on run time reasoning and inference (Kogut et al. 2002). Loosely speakin g, when adopted in computer science and artificial intelligence, an ontology is a world view, a means of viewing, organizing, conceptualizing and defining a domain of interest (Bergman 2007). The 2007 Ontology Summit summarized more than 40 terms that coul d be categorized under ontology,
17 including taxonomies, thesaurus, glossaries, and folksonomies (social bookmarks or tags). In a more formally defined context, an ontology is a set of formal axiomatic statements used to create a model of a domain for inter pretation. Here an ontology is a special kind of information object or computational artifact, or a means to formally model the structure of a system (Beetz 2006). There are several widely cited formal definitions of ontology. The first formal definition definitions above and r this research (Guarino et al. 2009). e expressed in a formal language with strict syntax and semantics. Otherwise, the ontology will not be defined in an unambiguous way and implied definition is not means the ontology should be used to represent consensual knowledge that is widely accepted in a domain by different participants. Ontologies are currently being used in various information related areas, such as natural language processing, information integration and data fusion. It promises sharing and reuse of knowledge of a domain of interest through a common computational form, thus bridging the gap between people and computers. The
18 introduction of Semantic Web makes ontologi es even more interesting and useful. The World Wide Web was initially designed as a document system whose content is meant to be displayed by Web browsers and is only meaningful to human readers rather than to computers (Cardoso 2007). In the so called syn tactic Web we are familiar with, the data and information expressed on a web page is difficult for a computer to extract and understand, preventing further automated information processing (Berners Lee 2001). The Semantic Web project was initiated in the h ope of giving order and meaning to the unstructured information available on the Web by adding contextual semantic annotation information (i.e. metadata) to existing information. Ontologies are used to establish a common understanding of the terminology by adding semantic markups to the current web content. As a result, two major objectives of the Semantic Web are identified: knowledge sharing and automatic information management by software agents. Computer programs will have the intellectual ability to di scover, understand and process the data from diverse sources automatically without human interference. Ontology Structure Among the terms in the ontology definition discussed above, the relevant domain (Guarino et al. 2009), or in other words, a model of the world or system. The entities of a system could be organized into concepts and relations. The concepts would have a taxonomy hierarchy from more generalized superconcep ts to more specialized subconcepts. The relations hold between relevant concepts. A separation is made between the concept (the TBox) and the instance of the concept (the ABox). The ontology should be defined as a generalized, stable form of knowledge, or a collection of universal rules, and is independent of any single world
19 state. So the instance of the concepts, i.e. the specific individual entity in real world, may not be relevant and may not appear in an ontology (TBox) (Guarino et al. 2009). Ontology ABox is only used in specific contexts where the instances of the concepts need to be addressed. Two main categories are identified for combining diversified ontologies. The first one is centralized approach, normally with administrative or contractual e ffort imposed by some governing party. This approach is suitable for ontologies developed within enterprise or community boundary, and enjoys economy, efficiency and interoperability advantages. Certain industries or disciplines, like pharmaceuticals and b iology, is adopting this approach, mostly as a result of efforts of the trade associations. The other approach, namely federated approach, is more suitable for broader spectrum like Internet and Semantic Web. The goal of ontology integration in a federated approach is to achieve interoperability at the data or instance level without unacceptable loss of information or corruption of the semantics (Bergman 2007). Ontology Languages and Web Ontology Language (OWL) The theories of ontology are backed by mathem atical foundations and are implemented in various ontology languages, among which OWL is selected as the one in this study. Description Logics (DL) is seen as the well suited ontology languages. As a subset of first order predicate logics, DLs are a famil y of languages that are used to representing declarative knowledge. DL is the basis of several well known ontology languages including OWL. Other languages are also available for ontology specification like Frame Logic (F logic). The advantage of F logic i s that it could be used for both ontology modeling and applications building using the ontology (Angele 2009).
20 Extensible Markup Language (XML) is the syntactic foundation for OWL. As a W3C endorsed standard for document markup, XML defines a generic synt ax used to mark up data with simple, human readable tags. It provides a standard format for computer documents, with the possibility of truly cross platform data interoperability, yet this format is flexible enough to be customized for diverse domains. As a light weight representation for data and knowledge on the Web, Resource Description Framework (RDF) is the machine understandable semantic annotations recommended by W3C in 1999 for the Semantic Web. RDF is a system on how to locate and describe the con tent and function of Web resources. Each individual resource is linked to a Universal Resource Identifier (URI) as a definitive reference, which uniquely identifies and may contain the information of the concept mentioned in the tag. An RDF statement is in properties in the form of URI or literals. Predicate links the subject and object together in a certain relat ionship. Subject and predicate must be URI resources; object could be URI resource or literal. RDF statements link the individual resources to form rules that could be used in the reasoning of computer programs (Pan 2009). An example may be the predicate. Together this statement specifies a relationship between window and opening. RDF Schema (RDFS) adds a little more content (extended vocabulary and relations) to the RDF baseline. RDFS statements are also simply RDF triples, but RDFS has predefined resources like rdfs:Class, rdfs:Resources and rdf:Property, etc. In those
21 predefined resources, rdfs:subClassOf and rdfs:subPropertyOf can support class and property hierar chies. rdfs:domain and rdfs:range is used to denote the domain and opening are classes while the fill is a property. The domain of this property may be limited to cert ain building elements like window or door, and the range of this property may be limited to openings (or even further, only the openings in walls) only. RDF and RDFS are often seen as the canonical, universal data model for data transfer and representatio n (Bergman 2007). They are also the basis of the W3C ontology recommendation for Semantic Web, Web Ontology Language (OWL), the ontology modeling language we choose in this research. RDFS is seen as a first try to support expressing simple ontologies with RDF syntax. However, the expressivity of RDF and RDFS is limited (Antoniou & van Harmelen 2009). Web Ontology Language (OWL) is being proposed by W3C as the ontology language of the Semantic Web. An OWL ontology could be subdivided into two parts: syntax and semantics. OWL extends RDF and RDFS and uses the same XML based syntax as the primary and mandatory syntax specified by W3C, although other syntaxes have also been defined for readability, including abstract syntax and graphical syntax in Unified Model ing Language (UML) among others UML is a widely adopted modeling tool in software engineering community. Kogut (2002) discussed the adoption of UML in ontology engineering in detail. The semantics is the meaning of the ontologies. The same semantics coul d be expressed in different syntaxes (W3C 2009).
22 The current version of OWL is OWL 2 based on XSD 1.1, which was published in October 2009, with two alternative ways to assign meanings to ontologies: OWL 2 DL and OWL 2 Full. The difference between the two flavors is from the dilemma between the expressiveness and reasoning power of an ontology. Expressiveness refers to the extent and convenience an ontology can be used to describe domain semantics. Expressiveness will be limited by syntactic restrictions. For example, in OWL Full, everything, even a property, is a class, which is not permitted in OWL DL. The more expressive an ontology is, the less efficient it could be reasoned on. In OWL Full, it cannot be reasoned on at all. Ontologies are used to model the static domain knowledge instead of dynamic reasoning knowledge, which are the topic of rule languages. Various rule languages are suggested for reasoning to explore the implicit knowledge hidden in an ontology. Examples of rule languages include SWRL and RuleML, etc. Current Domain Ontology Research in Construction Industry The research on ontology became popular in the AI communities in the early 1990s (Studer 1998). Currently ontologies provide semantic foundation for many software applications (Kog ut et al. 2002). While most of the theoretical researches of ontology are on more general and abstract human common sense knowledge, the ontology research on concrete subjects and tangible things is another field calls for more attention (Bergman 2007). Th ose ontologies are referred to as domain ontologies, which define concepts, activities, objects and the relationships among elements within a certain domain. Ontology modeling is the process to model concepts and relationships in a specific field or domain into formal ontologies, i.e. in an ontology language with formal syntax. The construction industry is seeing more domain ontology research and
23 applications. Several sources have been explored; pilot projects have been implemented; and different ontology b uilding process models have been proposed. T hose researches are discussed in this section as follows. The ontology research in the construction industry could be roughly divided into three stages. The first one is before the year 2000, during which time t he use of term on artificial intelligence (AI). For example, Chinowsky and Reinschmidt (1995) tried to judge whether a design satisfied certain non numerical specifi cations based on design CAD files, which is an early representative research in generating qualitative conclusions based on quantitative and geometric information available in CAD models. After the year 2000 comes the second stage, in which ontology becam e a popular research topic in construction industry, roughly 10 years after it became popular in computer science. Construction industry information and knowledge management is among the first areas focusing on the development and application of industry w ide ontologies. Several projects in Europe have addressed this problem. The construction knowledge management (KM) platform e COGNOS project is initiated for model based adaptive mechanisms that can organize documents according to contexts and interdepende ncies. A construction domain ontology is relied as a basis for knowledge indexing, discovery, and retrieval. The feasibility of using IFC for ontology building is also confirmed. This KM platform is built on the Web services model. Information is given on the development of the platform, but detailed information available for building the working ontology for the platform is scarce in the literature (Wetherill et al. 2002; Rezgui 2006). This field is still under research currently. For example, Wang (2010)
24 made use of ontology to represent and reason on context sensitive construction information as an alternative way of construction information management. Ontology is also used in product modeling. For example, Beetz (2006) used ontology in a topological rea soning service to deduce the area of a zone from CAD files (note that in a BIM model this information may already be available). The Geographic Information Systems (GIS) community is the field that pioneered ontology research in the AEC industry, as they n eed to process large amount of data. Many researchers, including Fonseca, et al. (1999), Nolan, et al. (2001) have noticed the key role of ontology in information integration and interoperability on semantic level. In 2006, Open Geospatial Consortium ( OGC ) examined the feasibility of representing Geography Markup Language (GML) in OWL as part of the preliminary effort to extend existing services, encodings and architectures with Semantic Web technologies (Akinci et al. 2008). Different possible sources have been explored as potential ontology sources for the construction industry. Besides more structured documents like specifications and OmniClass, less structured construction documents like OSHA safety recommendations have also been explored as a source of ontology (Wang 2010). Building codes also seem to be a promising alternative (Cheng et al. 2008). More ontology research in the construction domain means more overlapping but diverse ontologies available to choose from for a specific problem. When several ontologies are available in one domain, there may be two options to deal with. The first one is to integrate the ontologies into one. The other one is to choose one of them to use for a specific call. Semantic and ontology matching and mapping is becoming an
25 interesting topic since it plays an important role in joining heterogeneous ontologies to work together (Paolucci et al. 2002; Cheng et al. 2008). The third stage is about ontology research on BIM. Studies are being undertaken to investigate the oppor tunities to leverage the current IFC model to derive ontologies and develop standard models of the knowledge within the domain. The ISTforCE explored the development of an ontology to decode IFC models (Katranuschkov et al. 2003). They claimed that ontolog y is the key to human centered application supporting the engineer with access to the information on the project model. The ontology framework proposed serves as an advanced user gateway to product model data. They have provided rationale, principal design and technical structure of the framework, which is a good source of reference in building such a system. Although the framework is supposed to access IFC model data, the framework structure itself has little relationship with IFC specifications. Building such a framework from scratch is not justified considering the amount of effort being undertaken in the IFC and the fact that most of the work is overlapped. In the model checking framework proposed by Hjelseth and Nisbet (2010), ontology is treated as th e building block of knowledge and the implementation basis of the meta model. IfcOWL (Beetz et al. 2009) is by far the most complete effort to lift the IFC specification onto ontology level. However, since most of the ontology elements generated are strict ly rooted in IFC specification, its flexibility in different application scenarios might be restricted. Ontology research in BIM focuses on the physical or logical interdependency or hierarchy between objects and properties. It provides critical basic info rmation for
26 automated applications to reason through the objects. Although artificial intelligence cture of ontology, is recognized as a field lacks sufficient research (Jung and Joo 2010). Ontology alone may not prove very useful in the real world. Although Web services alone does not necessarily require ontology support, many researchers have resorte d to Web services to exert the power of ontology. Such ontology enhanced Web services are referred to as Semantic Web services. Issa and Mutis (2006) proposed a Semantic Web framework to address the reconciliation on different construction ontologies. Weth erill et al. (2002) suggested a knowledge management platform based on the Web the lifecycle information integrat ion of a building facility is also an agent based service oriented system. The effort of Vacharasintopchai et al. (2007) to build a working Semantic Web Services framework for computational mechanics is a good example of combining the Semantic Web and Web services together to work in the real world outside academic laboratories. Their framework is built on smart phone rather than normal desktop operating systems, which is very promising for mobile computing requirements such as on a construction jobsite. Bu t the detail about how the ontology is used in this system is not specified in the paper.
27 CHAPTER 3 DEVELOPMENT OF IFC BASED CONSTRUCTION I NDUSTRY ONTOLOGY Scope Definition According to Studer (1998), building a domain ontology should start with an ana lysis of the domain knowledge, resulting a taxonomy of concepts with attributes, values and relations. Additional information is captured in axioms. While it is possible to start an ontology from scratch, most of the ontology building process starts with s ome formal and formal knowledge (Sure et al. 2009). These contents usually already have some meaningful structure and organization, with the effort of ontology engineering these contents could be extracted and reus ed in a formal ontology for inference and data exchange. The construction industry ontology proposed in this paper is build based on the IFC specifications discussed in the previous section. Advantages of using IFC as an ontology source include modular de sign with different domains organized into different subparts and widely agreement on in the AEC community. The ontology is expected to reuse the rich content of currently developed IFC specifications and could be readily available to other construction ap plications to exploit the IFC content without involving complex techni cal issues of IFC. The content of IFC specifications is very broad, in the current stage of this research we restrict our ontology scope to the 3D modeling components part. According to Corcho (2002), an ontology should include the following minimal set of components: classes or concepts (with attributes describing the class), and relations or associations between concepts. Attributes and binary relations should be distinguished. Attribu tes are represented by basic data types, such as a number or a
28 describes the relat ion between the window and the wall. In an ontology the attributes are often expressed in datatype properties while the relations are expressed in object properties. Although the newest version of IFC specification is IFC2x4 RC2, the ontology source chose n for this research is the newest stable version of IFC specifications IFC2x3. This version is stable and widely used, and there is a corresponding XML schema publicly available. In addition, the core section of the IFC specification the 3D building elem ent breakdown, the spatial structure and the shape representation is well established and minimally changed (Liebich 2010). The OWL DL is the target ontology format, because it balances expressiveness and reasoning power of an ontology. Protg is used a s the ontology modeling tool (Knublauch 2004) Studer et al. (1998) argue that for a domain model to be qualified as an ontology, two additional conditions is required: generality and reusability level should be distinguished, and common understanding in a domain should be reached. Since t h e se factors should be discussed in specific application contexts and are beyond a technology problem, we do not include them in this research. A Closer Look on IFC Specification Structure The IFC specification is compos ed of detailed schema definition of hundreds of entities, e.g. IfcWall, IfcDoor, IfcWindow, etc. Those entities not only include concrete elements found in buildings, but also abstract concepts. Some of the abstract concepts are used to organize the entiti es, like IfcRoot and IfcObject; some are related to the process of building construction, like IfcProject and IfcOrganization.
29 The entities are organized into four layers: core shared, domain, and resource, as shown in F ig ure 3 1 The entities in the res ource layer are used to facilitate the description of entities in other layers and cannot appear independently, e.g. IfcLengthMeasure. All other entities are defined as children of IfcRoot, which is defined in IfcKernel schema. The entities forms a very co mplicated inheritance tree, in which those near the root are more abstract and vague, and those near the leaves are more specific and detailed. For example, IfcWindow is defined in the shared building element schema, which is in the shared layer. The IfcWi ndow specification defines the properties of a generic window. More details about the window lining or window panel are found in the architectural domain schema which is in the domain layer. IfcProduct, which is an ancestor of IfcWindow, is defined in the product ext ension schema in the core layer IfcProduct defines the properties that all the building products share, like physical location and geometric shape. Basic Ontology Development The basic ontology is the part of ontology components that can be d erived from the IFC specifications directly. The contents of the IFC specifications could fulfill most of the ontology components requirements, and forms the basics of the whole ontology. The IfcWindow specification is used here as an example. IfcWindow is a typical entity in IFC specifications. The specification page of IfcWindow includes the following sections: summary/definition, property set use definition, geometry use definitions, EXPRESS specification, attribute definitions and inheritance graph (b uildingSmart 2010 a ). The sections contained in other IFC elements are different due to the nature of each element, but are all structurally similar. The information contained in these
30 sections fits into the different components required for having an ontol ogy, as specified below. Figure 3 1 IFC2x4 Architecture Diagram (Source: buildingSmart (2010). "Industry Foundation Classes release 2x4 (IFC2x4) release candidate 2." < http://www. iai tech.org/ifc/IFC2x4/rc2/html/index.htm > (Mar. 15, 2010) ) The classes or concepts requirement of an ontology is about the nature or definition of certain terminology. They are also known as entities (this is the term used
31 a window, including the definition from ISO and IAI as well as an explanation of other IFC entities or types used by or related to IfcWindow. The Uniform Resource Locator (URL) of the webpage c ould be used as the URI to identify the term in the ontology. Classes are usually organized in taxonomies with inheritance information. This inheritance relationship back to the abstract entity IfcRoot, the ancestor of all independent IFC entities. The inheritance could be expressed by a subclass in the ontology. Table 3 1 shows the inheritance relationship from IfcRoot to IfcWindow. The (abs) after the entity name indicates t he entity is an abstract entity. Figure 3 2 shows the corresponding class hierarchy of the resulting ontology as shown in Protg. Table 3 1 Inh eritance relation for IfcWindow IFC Entity IFC Schema IfcRoot (abs) IfcObjectDefinition (abs) IfcObject (a bs) IfcProduct (abs) Core Kernel IfcElement (abs) IfcBuildingElement (abs) Core Product Extension IfcWindow IfcWindowStandardCase Shared Shared Building Elements In IFC, the attributes of the class is included in the following sections: proper ty set use definition, geometry use definitions and attribute definitions. Property sets are most typical attributes information. A property set is a group of properties that applies to each entity. The IfcWindow entity has three property sets: WindowCommo n, DoorWindowGlazingType and DoorWindowShadingType. The WindowCommon
32 property set includes reference, acoustic rating, fire rating, etc. Each property is described in a word (string) or a number, which are referred to as IfcPropertySingleValue in IFC. The other two property sets are similar properties about the glazing and shading of the window, but they also apply to doors. The reason to divide all these properties into three groups is to promote the re use of each property set through different entities. Other sections are also sources of entity attributes, for width of the window, defined in OverallHeight and OverallWidth attributes, with each value represented as a posit ive number, as shown in Figure 3 3 Figure 3 2. The class hierarchy in the ontology Figure 3 3 The properties of IfcWindow (Source: buildingSmart (2010). "IfcWindow." < http://www.iai tech.org/ifc/IFC2x4/beta3/html/ifcsharedbldgelements/lexical/ifcwindow.htm > (May 20, 2010) )
33 The relations in an ontology are also called roles. They denote how the classes or entities are associated with other cl asses. Most of the relations are binary, meaning two classes are involved. In IFC, most of the relations are defined as the subclasses of IfcRelationship, with prefix IfcRel. IfcRelationship is an abstract entity inherited from IfcRoot, on the same level o f IfcObjectDefinition shown in the table above. The relations between the IfcWindow class and other classes are described in the Figure 3 4 shows the relations a window (IfcRelContainedInSpatialStructure) in a buildin g story. The building itself is an Figure 3 4 The relations of IfcWindow (Source: buildingSmart (2010). "IfcWindow." < http://www.iai tech.org/ifc/IFC2x4/beta3/html/ifcsharedbldgelements/lexical/ifcwindow.htm > (May 20, 2010) )
34 Figure 3 5 shows the UML format of the partial sample ontology about IfcWindow. It is a subclass of IfcBuildingElement an d superclass of IfcWindowStandardCase. It has four properties or attributes, including simple ones like width and height expressed in simple numeric measurements and complex properties defined by other IFC elements. The same ontology is also expressed in R DF/XML format, as shown in Figure 3 6 Figure 3 5 A sample IfcWindow ontology in UML Figure 3 6 A sample IfcWindow ontology in RDF/XML
35 Extended Ontology Development Extended ontology is the ontology components that are not originally included in the IFC specifications but are added according to the requirement of the specific system or requirement. Since the purpose of the specific ontology we are using is information retrieval from an IFC model, some of the required ontology components may be out of the scope of the original IFC specifications. One of the extensions made is to facilitate the information retrieval when the input query includes terminologies that are not readily available in the current IFC specifications. For example, if the in put query includes a the new class (girder) and existing IFC classes (IfcBeam), and will not affect the hierarchical structure of the ontology. An equivalency between different classes could be rather complex with the boolean operations between different classes. For example if an ontology is developed will identify a nearby window as the most eff icient evacuation route instead a remote door with an exit tag on it. relation, which can be mapped to several IFC relationships, including the above mentioned IfcRelAggreg ates, IfcRelContainedInSpatialStructure, and others. For
36 kinds of relationship is trivial for the purpose of partial model retrieval, both of the the ontology. Other relations that do not specify a containment relationship is categorized as subrelationships the coding of specific requirement very concise compared to enumerating all the possible situations literally. While the basic ontology remains stable with each release of the IFC specifications, the e xtended ontology could be more versatile and updated more frequently according to the specific requirements of the different systems that the ontology is being used for.
37 CHAPTER 4 ONTOLOGY BASED INFORMATION RETRIEVAL FROM BUILDING INFORMATION MODEL S A Simple Ontology based Information Retrieval Use Case A simple illustration scenario of the lack of interoperability is the code checking of a building model against the building code. As most of the building codes have already been released online, they are available to be retrieved by software applications automatically. On the other hand, Building Information Models (BIM) complying to Industry Foundation Classes (IFC) specification are becoming a standard submittal requirement from owners. Technically, it would be possible to check the model against the building codes, but the reality is far from perfect. Among other issues, term matching is a problem. The information stored in the model is under the IFC naming convention, e.g. a window is under the elem enable the model and the building code to talk to each other, a translator is needed. The technique adopted by the model checking software applications currently is to use a hard coded dictionary, in which all the terms in building code are linked to the corresponding ones in IFC. This approach, while feasible, bears two major drawbacks. First, it is not easy to maintain the dictiona ry customarily, for example, adding new terminologies specific to a user environment. As the users of the software may lack the necessary knowledge of IFC specifications and/or code checking terminologies, it would be difficult for the user to add new spec ific terms into the dictionary when needed. It would be even worse if some special computer programming language is involved, e.g. EXPRESS, the native language used in IFC. Secondly, since each copy of the
38 dictionary is locally stored with the software on one computer, once an update is done it is difficult to spread the update to other computers and be utilized by the other users. The Partial Model Extraction Algorithm Although an IFC/ifcXML file is text based and could be easily opened in any text editor simply copying and pasting a chunk of the file to create a new one is not enough to make it useful, because the elements in an IFC file may refer to many other elements and/or be referred to by other elements. For example, the following line is the repre sentation of an IfcWindow element in an IFC file: #281=IFCWINDOW('0_p6ZzFwjAovJ0NxnsEEW_',#42,'Fixed:36" x 48":36" x 48":157225',$,'36" x 48"',#280,#274,'157225',4.,2.999999999999999); There are some explicitly described properties, including strings for its Global Unique ID object is also referring to at least three other elements, name ly #42, #280, #274, which might again refer to other elements. These referred elements may or may not include import information about this element. For example, #42 is IFCOWNERHISTORY, which is not really important, but #280 IFCLOCALPLACEMENT includes cri tical information on the exact location of the window. Besides, there are other syntax requirements (e.g. header information) for an IFC file to be valid. Our proposed approach is to extract a part from an IFC file while keeping its integrity using the IF C based ontology, which stores the information about possible relationships between each elements as well as necessary components for a valid IFC file as a whole.
39 In formal DL, there is a separation between an ontology of the axioms defining the classes a nd relations (TBox), and an ontology of the axioms of the individuals (ABox). The ontology developed according to Chapter 3 is an ontology TBox without any information of specific IFC elements in a concrete model. When the IFC file is read into the system, it is processed against the IFC ontology to generate an ontology augmented IFC index file, which is essentially an ontology ABox with all the IFC elements represented as ontology individuals. Meanwhile, a tree structure is also generated in the internal s torage format, with the IfcProject element as the root, or level 1. All the elements with a containment relation specified in the ontology with level n elements are structured as level n+1 nodes, or the children of level n nodes in the element tree. For ex basic IFC elements that cannot hold or contain other elements are always placed as th e not apply, but other relationship may still link them together. The interpreted tree structure is shown in Figure 4 1 he elements that go into the building. Figure 4 1 Tree structure of IFC elements
40 While this tree structure looks similar to the ones in normal IFC viewer as shown in Figure 4 2 a key difference is that the properties or relations specified in the IF C ontology is used to enhance the accuracy of the tree hierarchy. For example, on storey 2 shown in Figure 4 2 window is placed is categorized under storey 1, which may be the result that du ring the modeling process the whole external wall is modeled as one single wall. With the aid of the ontology, the system will know that since the window on storey 2 is connected with the wall, the wall should also appear under storey 2. Figure 4 2 Tre e structure of general IFC viewer s Figure 4 3 is an ontology enhanced tree view of the same model, please note that several IFC elements that are not visible under normal view is added under the building
41 storey. The ontology also stores the information o f the relations between the element individuals. Figure 4 3 Ontology enhanced tree structure of IFC elements Based on the ontology enhanced tree structure, a two pass partial model extraction algorithm is developed. The algorithm is described wi th th e example shown in Figure 4 4 The partial model extraction starts with some kind of location information as an input parameter. The location could be specified by an element directly, like a specific window. Or, it could also be in the form of the grid li nes if the grid system information is available in the model, in which case a comparison between IfcGrid and relevant building element is made to pin the desired location. The location will finally be
42 expressed by a specific element in the tree structure a s the starting point of the extraction. This will be a leaf node in the element tree, which is identified by its GUID. In the example, the window on the second floor is passed as the input parameter. Another input parameter would be used to specify how lar ge a partial model is needed, or the range of extraction, i.e. a whole floor or a room or simply a wall. The default value of the parameter would include all the objects that are immediately next to the specified element. Figure 4 4 Partial model ext raction sample The first pass is going up the decision tree in Figure 4 1 starting from the element to locate a proper container that could hold all the elements required for the partial model. This pass usually ends up at one of the four subclasses of I fcSpatialStructureElement, namely: IfcSite, IfcBuilding, IfcBuildingStorey, or IfcSpace. This pass in our sample model will end up with second storey IfcBuildingStorey.
43 After the first pass, the second pass is going down the tree from this container eleme nt to traverse all the potential elements. The elements that are connected with the starting element and other elements under the same container are checked. The location of each element is compared with the starting element. If the distance between the tw o elements is in a certain range specified as the second parameter of the algorithm, it will be selected for inclusion into the partial mode. According to the way the model is built, the relationship between the building elements might be different. For a n architecture model a single wall is usually built from level 1 up to the roof, while in a construction model the same wall may be divided into several horizontal sections with one section for each floor. Without loss of generality, we assume that the win dow actually is connected with the wall located on the first floor which is found in the first floor IfcBuildingStorey container. But the preprocessing of the model index file should already have identified the multiple location information of the wall, as discusses above. Next, all the selected elements are reassembled. At this point, some of the elements like the wall are reconstructed. Their shape is redefined and customized according to the extraction range input, and their position is adjusted to a new reference level. As a result, the resulting wall has a height of 10 feet instead of the original 20 feet. The tailored wall is actually placed on the second floor in the partial model, instead of the first floor on which it is placed in the original model If a new element is identified and included during this process, this will trigger another two pass examination as described above and more elements may be selected, until no more new elements are found. Finally, all the selected elements are reassembled into a new partial model.
44 The Prototype Application As mentioned in Section 2.1, Open IFC Tools is one of the third party APIs developed to assist the access of information stored in an IFC STEP file. The API is implemented in Java programming language a nd is developed at Bauhaus University of Weimar, Germany. It works as an interface between the Java program and the native IFC file so one can call the abstracted methods and the API will finish corresponding operations on the lower level. T he ontology developed in Chapter 3 and the extraction algorithm designed in Section 4.1 are further implemented in a Java prototype application to demonstrate the use of the ontology in information retrieval and partial model extraction process. The system a rchitectur e is shown in Figure 4 5 Figure 4 5 System architecture of the application
45 The main system could be divided into three subparts. The first part is the mapping between the original IFC file and the generated ontology augmented IFC index file. When a new IFC file is read into the system, it is processed against the IFC ontology TBox developed in Chapter 3 to generate an ontology augmented IFC index file. The IFC ontology TBox only has classes and properties defined. During the processing each IFC eleme nt in the IFC file generates an ontology individual of the corresponding class in the IFC ontology model, and the resulting ontology ABox with individuals added forms the IFC index file. In the above mentioned extraction example, in the general IFC ontolog y TBox only one IfcWindow class is defined A fter the process each of the two windows generates a specific IfcWindow individual (or instance) in the ontology model. The individuals keep reference mapping back to the original elements in the IFC file with t heir GUIDs. As a result, those IFC elements have not only the relations directly defined in the IFC model, but also those defined in the ontology. The second part is the manipulation on the IFC files. Adopted in our system is the OpenIFCTools API (openifc tools 2011), which is a Java Object Oriented implementation that works on the native IFC format files directly. The third part is the query on the ontology augmented IFC model, which is implemented with Jena API (Jena 2011), which is an API for working wit h ontologies and Semantic Web applications. Figure 4 6 shows a screenshot of the user interface.
46 Figure 4 6 Screenshot of the prototype Java application
47 CHAPTER 5 CONCLUSION AND RECOMMENDATIONS By combining the strength of ontology and IFC techno logies, this study explores the possibility of extracting a partial model from an IFC model with the help of information indexed by an IFC based ontology. A sample ontology is developed and implemented in a prototype Java application. It is shown that usin g IFC specifications to build a construction industry ontology is a valid way to reuse the domain knowledge from IFC effort and to exploit the IFC building information models. As an alternative way to utilize the information in a BIM model, the ontology b ased partial model extraction approach described in this study is one of the many possibilities for utilizing a construction industry domain ontology. As mentioned in Katranuschkov et al. (2003), a properly developed domain ontology could hide the model co mplexity, hence separating the more complicated process of knowledge building and the process of using the knowledge stored in an ontology. Further research and exploration on both the ontology and the application framework are expected. The current sampl e ontology only covers a small fraction of IFC elements. A complete industry wide ontology is far more complicated and requires much more work to build and constant maintenance Studer (1998) has noticed the similarity in the difficulty of scaling up an ac ademic prototype t o large knowledge based systems in real world is waiting for further validation. Currently the ontology TBox and the augmented ontology ABox are both stored into the computer memory at run time. This flexible way of ontology handling is suitable for small to medium size models, but when the model is larger, this method may cause
48 significant overhead on loading the model. If the model is larger than the memory part of it will be forced to be stored in secondary memory like disks, which will cause even more obvious overhead on ontology operations. Jena has a persistent on tology model that could handle ontology models stored in a database system. This might be a solution for applying this framework on big building models that has tens of thousands of ontology instances in an augmented model. Besides the stand alone prototype application, a standard Web services is expected to be implemented so that the service co uld be connected to by other applications via Web, and possibly be used in a cloud computing framework. For example, every morning the 4D scheduling software could be configured to use the scheduling information to extract the partial model that the team i s going to work on for that day from the complete model located at an off site location, and it will have the partial model ready for the kickoff meeting before the crew arrives at work.
49 LIST OF REFERENCES Adachi, Y. (2003). "Overview of partial model q uery language." 10th International Conference on Concurrent Engineering Madeira, Portugal. Akinci, B., Karimi, H., Pradhan, A., Wu, C. C., and Fichtl, G. (2008). "CAD and GIS interoperability through semantic web services." Journal of Information Technolo gy in Construction 13, 39 55. Antoniou, G., and van Harmelen, F. (2009). "Web ontology language: OWL." Handbook on Ontologies S. Staab, and R. Studer, eds., Springer, Berlin, Germany, 91 110. Beetz, J., de Laat, R., van Berlo, L., and van den Helm, P. (2 010). "Towards an open building information model server." 10th International Conference on Design and Decision Support Systems in Architecture and Urban Planning The Netherlands Beetz, J., van Leeuwen, J., and de Vries, B. (2006). "Towards a topological reasoning service for IFC based building information models in a semantic Web context." Joint International Conference on Computing and Decision Making in Civil and Building Engineering Montreal, Canada, 3426 3435. Beetz, J., van Leeuwen, J., and de Vrie s, B. (2009). "IfcOWL: A case of transforming EXPRESS schemas into ontologies." Artificial Intelligence for Engineering, Analysis and Manufacturing 23, 89 101. Bergman, M. K. (2007). "An intrepid guide to ontologies." < http://www.mkbergman.com/374/an intrepid guide to ontologies/ > (Mar. 15, 2010). Berners Lee, T., Hendler, J., and Lassila, O. (2001). "The semantic Web." Scientific American 284(5), 34. Brandon, P. S., and Betts, M. (1995). Integrated construction information Taylor & Francis. buildingSmart (2010 a ). "IfcWindow." < http://www.iai tech.org/ifc/IFC2x4/beta3/html/ ifcsharedbldgelements/lexical/ifcwindow.htm > (May 20, 2010). buildingSmart (2010 b ). "Industry Foundation Classes release 2x4 (IFC2x4) release candidate 2." < http://www.iai tech.org/ifc/ IFC2x4/rc2/html/index.htm > (Mar. 15, 2010). buildingSmart (2011). "Model View Definition summary." < http://buildingsmart tech.org/specifications/ifc view definition/s ummary > (Jul. 15, 2011). Campbell, D. A. (2007). "Building Information Modeling: the Web3D application for AEC." 12th International Conference on 3D Web Technology New York, NY, 173.
50 Cardoso, J. (2007). Semantic Web services: theory, tools and applicatio ns IGI Global, Portugal. Cheng, C. P., Lau, G. T., Pan, J., Law, K. H., and Jones, A. (2008). "Domain specific ontology mapping by corpus based semantic similarity." 2008 NSF CMMI Engineering Research and Innovation Conference Knoxville, Tennessee. Chino wsky, P. S., and Reinschmidt, K. F. (1995). "Qualitative geometric reasoner for integrated design." Journal of Computing in Civil Engineering 9(4), 250 258. Corcho, O., and Fernandez Lopez, M. (2002). "Ontological engineering: what are ontologies and how can we build them?" Semantic Web services: theory, tools and applications J. Cardoso, ed., IGI Global, Portugal. Dickinson, I. (2009). "Jena ontology API." < http://jena.sourceforge.net/ontolo gy/index.html > (Jul. 4, 2011). Eastman, C., Teicholz, P., Sacks, R., and Liston, K. (2008). BIM handbook: a guide to Building Information Modeling for owners, managers, designers, engineers, and contractors Wiley, Hoboken, New Jersey. Fernandez Lopez, M. (1999). "Overview of methodologies for building ontologies." 16th International Joint Conference on Artifical Intelligence Stockholm, Sweden. Fonseca, F., Egenhofer, M., and Davis, C. (1999). "Ontology driven information integration." AAAI Technical Repo rt WS 00 08 AAAI, Austin, TX. Gallaher, M. P., O'Connor, A. C., Dettbarn, J. L., and Gilday, L. T. (2004). "Cost Analysis of Inadequate Interoperability in the U.S. Capital Facilities Industry." NIST. Gruber, T. R. (1993). "A translation approach to porta ble ontology specifications." Knowledge Acquisition 5(2), 199 221. Guarino, N., Oberle, D., and Staab, S. (2009). "What is an ontology." Handbook on Ontologies S. Staab, and R. Studer, eds., Springer, Berlin, Germany, 1 17. Hjelseth, E. (2010). "Overview of concepts for model checking." 27th International Conference on Applications of IT in the AEC Industry and Accelerating BIM Research Workshop Cairo, Egypt. Issa, R. R. A., and Mutis, I. (2006). "Ontology based framework using a semantic Web for address ing semantic reconciliation in construction." International Workshop on Intelligent Computing in Engineering 348 367. Jena (2011). "Jena A Semantic Web framework for Java." < http://jena.sourceforge.net/ > (Ma r. 15, 2011).
51 Jung, Y., and Joo, M. (2011). "Building Information Modeling (BIM) framework for practical implementation." Automation in Construction 20, 126 133. Katranuschkov, P., Gehre, A., and Scherer, R. J. (2003). "An ontology framework to access IFC model data." Journal of Information Technology in Construction 8, 413. Katranuschkov, P., Weise, M., Windisch, R., Fuchs, S., and Scherer, R. J. (2010). "BIM based generation of multi model views." 27th International Conference on Applications of IT in t he AEC Industry and Accelerating BIM Research Workshop Cairo, Egypt. Knublauch, H., Fergerson, R. W., Noy, N. F., and Musen, M. A. (2004). "The Protege OWL plugin: an open development environment for semantic Web applications." 3rd International Semantic Web Conference Kogut, P., Cranefield, S., Hart, L., Dutra, M., Baclawski, K., Kokar, M., and Smith, J. (2002). "UML for ontology development." The Knowledge Engineering Review 17(1), 61 64. Lee, G., Eastman, C., and Sacks, R. (2006). "Generating IFC view s and conformance classes using GTPPM." Joint International Conference on Computing and Decision Making in Civil and Building Engineering Montreal, Canada, 1715 1724. Lee, J. S., Min, K. M., Lee, Y. S., Kim, J. H., and Kim, J. J. (2008). "Building ontolog y to implement the BIM focused on pre design stage." 25th International Symposium on Automation and Robotics in Construction Vilnius, Lithuania, 350 354. Liebich, T. (2010). "Unveiling IFC2x4 the next generation of openBIM." 27th International Conferenc e on Applications of IT in the AEC Industry and Accelerating BIM Research Workshop Cairo, Egypt. Nolan, J. J., Simon, R., and Sood, A. K. (2001). "Developing an ontology and ACL in an agent based GIS." Workshop on Ontologies in Agent Systems Montreal, Ca nada. OMG (2009). "Ontology definition metamodel." < http://www.omg.org/spec/ODM/ > (May 10, 2010). OpenIFCTools (2011). "Open IFC Tools." < http:// openifctools.com/Open_IFC_Tools/Home.html > (Jul. 1, 2011). Pan, J. Z. (2009). "Resource description framework." Handbook on Ontologies S. Staab, and R. Studer, eds., Springer, Berlin, Germany, 71 90. Paolucci, M., Kawamura, T., Payne, T., and Sycara, K. (2002). "Semantic matching of web services capabilities." 6th IEEE International Symposium on Wearable Computers Seattle, WA, 333.
52 Pathak, J. (2006). "Difference between Agents and (Semantic) Web Services." < http://lists.w3.org/Archives/Public/public sws ig/2006Oct/0074.html > (Mar. 15, 2011). Shen, W., Hao, Q., and Xue, Y. (2010). "An agent based service oriented approach for facility lifecycle information integration and decision supports." 2010 IEEE International Conference on Systems Management and Cybernetics 1945 1952. Stadler, A., and Kolbe, T. H. (2007). "Spatio semantic coherence in the integration of 3D city models." 5th International ISPRS Symposium on Spatial Da ta Quality Enschede, The Netherlands. Studer, R., Benjamins, V. R., and Fensel, D. (1998). "Knowledge engineering: principles and methods." Data & Knowledge Engineering 25(1 2), 161 197. Sure, Y., Staab, S., and Studer, R. (2009). "Ontology engineering m ethodology." Handbook on Ontologies S. Staab, and R. Studer, eds., Springer, Berlin, Germany, 135. Tizani, W., and Mawdesley, M. J. (2011). "Advances and challenges in computing in civil and building engineering." Advanced Engineering Informatics 25(4), 569 572. TNO (2010). "IFC Engine DLL." < http://www.ifcbrowser.com/ifcenginedll.html > (Mar. 15, 2010). Vacharasintopchai, T., Barry, W., Wuwongse, V., and Kanok Nukulchai, W. (2007). "Semantic Web services framework for computational mechanics." Journal of Computing in Civil Engineering 21(2), 65. Vanlande, R., Nicolle, C., and Cruz, C. (2008). "IFC and building lifecycle management." Automation in Construction 18, 70. W3C (2009). "OWL 2 Web Onto logy Language Primer." < http://www.w3.org/TR/2009/REC owl2 primer 20091027/ > (Mar. 15, 2011). Wang, H. H., and Boukamp, F. (2011). "Ontology based representation and reasoning framework f or supporting job hazard analysis." Journal of Computing in Civil Engineering 25(6), 442 456. Weise, M., Katranuschkov, P., and Scherer, R. J. (2003). "Generalised model subset definition schema." International Conference on Applications of IT in the AE C Industry Wetherill, M., Rezgui, Y., Lima, C., and Zarli, A. (2002). "Knowledge management for the construction industry: the e COGNOS project." Journal of Information Technology in Construction 7, 183.
53 Zhang, J. (2010). "A social semantic Web system fo r coordinating communication in the architecture, engineering and construction industry." Ph.D. dissertation, University of Toronto, Toronto, Canada.
54 BIOGRAPHICAL SKETCH Le Zhang is a scholar with broad interest from China. He was admitted into the c ons truction p roject m anagement program of Tianjin University, Tianjin, China in 2000, where he earned his Bachelor of Management degree. During his undergrad study, he also finished a second degree of Bachelor of Art s in English. In 2005 h e continued his stud y in international construction contract document s in Tianjin University and earned his Master of Management degree. He was admitted into the Ph.D. program in M.E. Rinker Sr. School of Building Construction, University of Florida, Florida, US in 2007 and started research on Building Information Modeling (BIM). H e also pursued a non traditional concurrent degree program with a Master of Science in c omput er e ngineering in University of Florida