past, present and future of information and knowledge sharing in the construction industry: towards...

Computer-Aided Design 43 (2011) 502–515

Contents lists available at ScienceDirect

Computer-Aided Design

journal homepage: www.elsevier.com/locate/cad

Past, present and future of information and knowledge sharing in theconstruction industry: Towards semantic service-based e-construction?Yacine Rezgui a,∗, Stefan Boddy b, MatthewWetherill b, Grahame Cooper ca Cardiff University, School of Engineering, Queen’s Buildings, The Parade, Cardiff, CF24 3AA, Wales, UKb Salford Centre for Research and Innovation in the Built and Human Environment, Maxwell Building, University of Salford, The Crescent, Salford, Greater Manchester, M5 4WT, UKc Informatics Research Institute, Maxwell Building, University of Salford, The Crescent, Salford, Greater Manchester, M5 4WT, UK

a r t i c l e i n f o

Article history:Received 8 August 2007Accepted 11 June 2009

Keywords:Product data technologySTEP (ISO 10303)IFCBIMOntologyCADWeb servicee-Process

a b s t r a c t

The paper reviews product data technology initiatives in the construction sector and provides a synthesisof related ICT industry needs. A comparison between (a) the data centric characteristics of ProductData Technology (PDT) and (b) ontology with a focus on semantics, is given, highlighting the pros andcons of each approach. The paper advocates the migration from data-centric application integration toontology-based business process support, and proposes inter-enterprise collaboration architectures andframeworks based on semantic services, underpinnedbyontology-based knowledge structures. Thepaperdiscusses the main reasons behind the low industry take up of product data technology, and proposes apreliminary roadmap for the wide industry diffusion of the proposed approach. In this respect, the paperstresses the value of adopting alliance-based modes of operation.

© 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Construction is a knowledge intensive industry characterizedby its unique work settings and virtual organization like modusoperandi [1]. The Construction sector is fragmented and the majorconsequence is the difficulty to communicate effectively andefficiently among partners during a building project or betweenclients and suppliers of construction products. Several initiativesled by standardisation and/or industry consortia have developeddata/product models aimed at facilitating data and informationexchange between software applications. These efforts includeSTEP [2] and the Industry Foundation Classes (IFCs) [3]. Severalother initiatives at a national and European level have developeddictionaries, thesauri, and several linguistic resources focusedon Construction terms to facilitate communication and improveunderstanding between the various stakeholders operating ona project or across the product supply chain. However, theseinitiatives tend to be country specific and not adapted to themulti-national nature of the sector. Also, given the vast scopeof Construction, these semantic resources tend to be specializedfor dedicated applications or engineering functions, e.g. productlibraries and HVAC (Heating, Ventilation and Air Conditioning),respectively.

∗ Corresponding author. Tel.: +44 2920 875719.E-mail address: [email protected] (Y. Rezgui).

0010-4485/$ – see front matter© 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.cad.2009.06.005

A comprehensive literature review targeting Computer Inte-grated Construction (CIC) was reported by the authors in [4]. Thisreview reveals a strong focus on data and application integra-tion research. It is argued that whilst valuable, such research andthe software solutions it yields fall short of the potential for CIC.Thus the authors call for re-focussing CIC research on the rela-tively under-represented area of semantically described and co-ordinated process oriented systems to better support the kind ofshort term virtual organisation that typifies the working environ-ment in the construction sector. Moreover, the review providesa Framework that illustrates the CIC research landscape (Fig. 1).A two dimensional representation is used, developed with re-spect to two axes: (a) Semantic Focus— this axis spans the wholespectrum of past, existing, and future applications with under-lying semantics ranging from data structures conveyed throughdata models to rich-semantic representations through ontology;(b) Application Domain Focus — this axis represents the focus ofresearch effort on a continuum from application and API (Applica-tion Programming Interface) centric, to process and people centric.

This paper focuses on the right-bottom quadrant of Fig. 1 andargues the case for a change of emphasis from data and objectcentric applications to high-level process driven semantic services.It builds on the results of a wide consultation led by the authors inthe context of the EU funded ROADCON project [5] that resulted in(a) comprehensive industry requirements, (b) an ICT vision, and(c) the first ICT roadmap for the Construction industry. In fact,the authors’ research over the last decade (as illustrated in Fig. 2)

http://dx.doi.org/10.1016/j.cad.2009.06.005

http://www.elsevier.com/locate/cad

http://www.elsevier.com/locate/cad

mailto:[email protected]

http://dx.doi.org/10.1016/j.cad.2009.06.005

Y. Rezgui et al. / Computer-Aided Design 43 (2011) 502–515 503

Fig. 1. Computer Integrated Construction research landscape [4].

Fig. 2. Authors’ involvement in national and EU CIC initiatives.

has evolved from advanced data and information managementsolutions [6], applied later in the context of CAD [7], to advancedknowledge management systems, articulated around the use of anontology [8,9], and deployed in distributed environments [10,11].Moreover, the emphasis on knowledge infused applications usingservice-oriented architectures has been made and reported in[12–15].

As such, the papermakes fourmain contributions: (a) it outlinesand discusses the various approaches to integration adopted by the

Construction IT community; (b) it argues the case for ontologiesas a means to address limitations faced by existing Product/Datamodel standards; (c) it puts ontologies into context by promotinga service-oriented view of the world; and (d) it discusses adoptionand diffusion issues informed by existing STEP/IFC initiatives andproposes a way forward.

The paper is organised as follows. First, the methodology thatunderpins the proposed research is presented in Section 2. Sec-tion 3 reviews two decades of product data research from early

504 Y. Rezgui et al. / Computer-Aided Design 43 (2011) 502–515

product models to current so called Building Information Models(BIM). A critical discussion on product data technology is then pro-vided in Section 4, arguing the case for knowledge-rich ontology.Section 5 discusses industry adoption and diffusion of IFCs, andidentifies key requirements for the development and adoption ofa construction ontology. Section 6 then discusses how an ontologycan play a pivotal role in enabling seamless inter-working and in-teroperability between diverse web-enabled applications, namelyweb-services, and identifies essential requirements for supportingdynamic, long lasting, processes as experienced during the designstage of a building. Section 7 provides a description of the techni-cal infrastructure followed, in Section 8, by proposals for a stagedroadmap, inspired from the ROADCON project [5], to help the in-dustry migrate from current product data centric applications tosemantic ontology-enabled services.

2. Methodology

The research presented in the paper adopts a reflective pra-ctice approach underpinned by a participatory action researchmethodology [16]. There are two sorts of reflection: reflection-in-action and reflection-on-action. The former involves a level ofawareness throughout the action, while in the latter reflection isdone after the action or event. The paper adopts a reflection-on-action approach. In fact, several case study projects that span aperiod of over ten years form the focus of this study. These aredepicted in Fig. 2. The value of a person’s reflection can be greatlyenhanced by a greater understanding of the process. This allowsthe participants to create their own knowledge and theory relevantto their own specific situation.

The paper will reflect on the authors’ 15 years active involve-ment in national (UK) and EU funded research (as illustrated inFig. 2) and propose a re-focussing of CIC research on high-level pro-cess driven semantic services.

3. From simple product data to complex building informationmodels

Although the manual referencing of paper based product dataand building design has existed for centuries, it was the increasinguse of CAD facilities in design offices from the early 1980swhich prompted the first efforts in electronic integration andsharing of building information and data [4]. Here, the abilityto share design data and drawings electronically through eitherproprietary drawing formats or via later de facto standards suchas DXF (Drawing/Data Exchange Format), together with the addeddimension of drawing layering had substantial impacts on businessprocesses and workflows in the construction industry [17,18].Although in these early efforts, sharing and integrationwasmainlylimited to geometrical information [19], effectively the use of CADfiles was evolving towards communicating information about abuilding in ways that a manually draughted or plotted drawingcould not [20].

This evolution continued with the introduction of object-oriented CAD in the early 1990s by companies such as AutoDesk,GraphiSoft, Bentley Systems etc. Data ‘‘objects’’ in these systems(doors, walls, windows, roofs, plant and equipment etc.) storednon-graphical data about a building and the third party compo-nents which it comprises ‘‘product data’’, in a logical structure to-getherwith the graphical representation of the building [21,20,22].These systems often supported geometrical modeling of the build-ing in three dimensions, which helped to automate many of thedraughting tasks required to produce engineering drawings.

When combined with the increasing ubiquity of electronicnetworking and the Internet, this allowed many companies tocollaborate and share building information and data which in

turn led to new ways of communicating and working [21,23,24].The opportunities presented by the move towards collaborativeworking and information sharing encouraged a number of re-search projects in the early 1990’s, which aimed to facilitate andprovide frameworks to encourage the migration from documentcentred approaches towards model based, integrated systems:CONDOR [25]; COMMIT [6] being examples. Similarly, the OS-MOS [10] research project aimed to develop a technical infrastruc-turewhich empowered the construction industry tomove towardsa computer, integrated approach.

It became clear that in order to take best advantage of thepotential for CAD and object/product model integration, there wasa need for more coordinated standards which would simplify andencourage its uptake [26]. These standards defining efforts camein the form of the STEP application protocols for construction[3,27,28]. This work, inspired by previous work primarily in theaerospace and automotive fields, formed part of ISO 10303, theInternational Standard for the Exchange of Product Model Data.Latterly, the International Alliance for Interoperability defined theIndustry Foundation Classes, a set of model constructs for thedescription of building elements. Preceding and in some casesconcurrent with this work, the research community producedseveral integratedmodel definitions including GARM [29], the AECBuilding Systems Model [30], ATLAS [31], the RATAS model [32],OPIS [33], and the COMBINE Integrated Data Model [3]. Theseresearch efforts tended to propose a data model and also providea suite of tools to manipulate the model (as a proof of concept), ora central database to serve model elements to other applicationsused in the construction project process via some form ofadapter [34–36]. One of themost recent incarnations of the centraldatabase idea can be seen in the IFC Model Server from VTT ofFinland [37] designed to host entire building models described inthe IAI IFC format.

Within the last three to four years, researchers and commer-cial applicationdevelopers in the constructiondomainhave startedto develop tools to manipulate complex building models [38,28].By storing andmanaging building information as databases, build-ing information modeling (BIM) solutions can capture, manage,and present data in ways that are appropriate for the user of thatdata. Because the information is stored in a logically centraliseddatabase, any changes in building information data can be logi-cally propagated andmanaged by software throughout the projectlife cycle [20,39]. Building information modeling solutions add themanagement of relationships between building components be-yond the object-level information in object-oriented CAD solu-tions. This allows information about design intent to be capturedin the design process. The building informationmodel contains notonly a list of building components and locations but also the rela-tionships that are intended between those objects [40,41].

This new wave of BIM applications, embody much of the visionof previous academic research such as ATLAS, and COMBINE,whilststill relying on data exchange standards or API level customisationfor interoperability/integration. Recently, the American NationalInstitute of Building Sciences has inaugurated a committee to lookinto creating a standard for lifecycle data modeling under the BIMbanner [8]. The idea here is to have a standard that identifies datarequirements at different lifecycle stages in order to allow a moreintelligent exchange of data between BIM enabled applications.

4. Product data versus knowledge-rich ontology

This section discusses first the philosophical underpinnings ofproduct data and ontology, and then provides a critical review oftheir shortcomings, arguing the case for an enhancement of prod-uct data to pave the way to more effective, user-friendly concep-tualisations of the construction domain through ontology.


4.1. Philosophical underpinnings of product data and ontologyapproaches

Various definitions of what forms an ontology have been for-mulated and have evolved over time. A good description of thesecan be found in [42]. From the authors’ perspective, the best def-inition that capture’s the essence of an ontology is the one givenby Gruber, [43]: ‘‘an ontology is a formal, explicit specification ofa shared conceptualization’’. As elaborated in [44]: ’’Conceptual-ization refers to an abstract model of some phenomenon in theworld which identifies the relevant concepts of that phenomenon.Explicit means that the types of concepts used and the constraintson their use are explicitly defined. Formal refers to the fact that theontology should be machine processable’’.

People often find it difficult to see clearly howan ‘‘ontology’’ dif-fers fromwhat they already recognize as a ‘‘data model’’, focussingon the formal nature and structuring mechanisms that seem to becharacteristic of both. Certainly, data modeling languages providethe ability to define taxonomies through notations that supportclassification, generalization and specialization, they support thedefinition of relationships or associations between concepts, andideas of aggregation and composition, and in terms of these prim-itives appear to offer the same support for representing conceptsand the relationships between them. However, the authors wouldmaintain that trying to understand the distinctions in terms of themodeling primitives that are used is a mistake; it is the nature ofthe models themselves, the way in which they are derived, andthe tools that support their use that provides the differentiation.In order to understand this, it is necessary to return to the under-lying problems thatmake it difficult to achieve a single agreed datamodel for an industry.

Returning to Gruber’s definition of an ontology, a key elementis the idea of a shared conceptualization [43]. Typically, in humanendeavour, shared conceptualizations are defined over a lengthyperiod of time, based on the shared experience of a group of people,sometimes referred to as a community of practice [45]. They willinvolve the definition and use of abstractions that are designedto capture the important aspects of some practical context inorder to support a particular activity or type of activity. As such,a shared conceptualization is a socially constructed model orreality that is distinct from reality and is optimized to supportthe goals and activities of the community of practice in which itwas defined. Communities engaged in different activities are likelyto form shared conceptualizations that are quite different viewsof reality, and make up shared ‘‘world-views’’ [46] that provide abasis for highly effective and efficient communications within therespective communities.

In order to understand and formalize the shared world-views of such communities in the form of ontologies to supportthe integration of diverse human activities, it is important toconsider approaches that derive from an interpretive philosophicalstandpoint rather than from a positivist, scientific/engineeringone [27]. In such an approach, we try to interpret, accommodateandmodelwhat is, rather than trying to change reality to fit a singlemodel. This inevitably results in different ontologies for differentcommunities, but the challenge then is to find ways to allow thosecommunities to collaborate effectively with one another whilstmaintaining their existing, efficient, effective separate world-views. The implication is that we are forced to shift our emphasisfrom developing a standard representation of a single ‘‘reality’’,towards providing mechanisms for supporting communicationbetween differing perceptions of reality, focussing our attention onthe overlaps at the boundaries and the specific conceptualizationsthat are required for such communication to happen.

An important consequence of this shift is that it becomespossible to adopt a more incremental approach to the integration

of processes across disciplines. The single data model approachcan tend to result in the need for an ‘‘all-or-nothing’’ approachto implementation, and certainly practical issues have been notedregarding the size and manageability of IFC models [47]. Evenwith the existence of product model servers [48], the practicalimplementation of single-model-based integration seems fraughtwith difficulty.

4.2. Alleviating Product Data shortcomings through ontology

The progress made so far in arriving at the BIM concept and itsassociated tools is undoubtedly a sizeable step forward in theman-agement, communication and leveraging of construction projectinformation. Both the BIMmodels used by the commercial vendorsand the international standards developed for construction such asSTEP, IFC and CIS/2 do however still exhibit shortcomings as high-lighted in [49,27] and from our own observations. These shortcom-ings are identified in Table 1 alongwith the level of support offeredby both the existing popular standards (including IFCs) and the po-tential ontology based solutions.

It is clear then that the ontology approach is by no meansa cure-all for the ills of product data models as they currentlystand. Indeed we would not propose to replace data models somuch as enhance them, and the applications used to create andmanipulate them, with additional semantic information basedon domain and core ontologies, as described in the followingsection. However, we would maintain that ontologies have theright interpretive philosophical underpinning that is more likelyto address the information and knowledge sharing requirementsof the construction user community as argued in Section 4.1.

The Web Ontology Language (OWL) is the current leadinginternational standard notation for the definition of ontologies in amachine interpretable fashion. OWL has two primary constructs,namely Classes and Properties. Classes represent categories ofthings, real or conceptual, and Properties define the relationshipsbetween Classes and between instances of those classes. An OWLClass is somewhat analogous to an Express Application Object andwhereas Express really only has explicit relationships of the ‘is a’inheritance hierarchy type, OWL Properties are far more flexibleand explicit in describing a richer set of possible relations betweenits classes. Many Express relationships are opaquely embeddedin the properties of application objects. Similarly, the Expressbased IFCs define a slightly broader range of relationships but stillsomewhat fewer than are routinely embedded in anOWLontology.It is this ability to define rich relationships between Classes thatgives an OWL encoded ontology its power.

We see other problems that render data level integration inthe STEP or IFC mould less effective than might otherwise be thecase. To understand this position it is necessary to consider theway in which data integration mandates a considerable degree ofwork up-front. This is required in order to agree upon standards,construct a schema for integration, adapt applications to thestandards etc., all before any benefits are realised. These issuesbecome more onerous the larger the scope of agreement oneis trying to achieve (inter-organisational, national, internationaletc.). Finally, for large international standards efforts, agility issomething of a problem. Once the standard is agreed, changing itcan take a considerable amount of time, which in an age of rapidlyevolving business needs can turn a formerly helpful system into ahindrance [4].

The use of an ontology or multiple ontologies of the construc-tion domain could act as a semantic abstraction layer above cur-rent standards and models to further integrate project data in amore intelligent fashion. For example, taking the point on viewsfrom Table 1, we believe an ontology with mappings into the un-derlying data models could be used to provide a more intuitive


Table 1Product data versus ontology.

Issue STEP/IFC Ontology

Information schema evolution through time tosupport changing project or industry contexts

Not fully catered for. The IFC Property Set construct could be employed tofulfil role for certain types of information.

This is an inherent attribute ofontology although not cateredfor specifically; but awell-maintained and updatedontology should evolve with thedomain quite naturally (whilstrespecting and allowing forbusiness processes which mayrequire a stable schemainterpretation).

Views on data aligned to user and application needs Views on STEP models can be defined in Express-X. The Application Protocolsare themselves domain specific views to some extent. Other work is ongoingto extend models into specific sub-domains (e.g. [50])

Base level domain ontologiescould be said to be views in theirown right as they support theinformation needs ofcommunities of practice. Thesemay be able to be transformedthrough the mechanism of anupper ontology to suit.

Object ownership and rights management Not supported, but can be via EXPRESS Not supported but can be viaOWL

Lifecycle management and placement of data in theprocess

STEP defines basic resources dedicated to relating data to its place in theprocess

An ontology of constructionwould include conceptsdescribing the constructionprocess and would thereforelikely feature relations to classesof data involved in the process.This would however be a lessfixed notion than the STEPapproach.

Recording/embedding of decision rationale Some basic support [51] No support

Links to external information – particularlyunstructured information

Linkscan be manually defined, but they have no specific semantics. Ontologies tend to be built fromtext documents usingInformation retrievaltechniques. The latter are usedto infer external information’s‘relation’ to ontology conceptsand by examining other existingrelations, its links to other data.

view of project data for any given actor based on their particulardisciplinary concepts and terminology. That same ontology couldalso provide the view for an actor from a different discipline, basedon the relationships explicated within the domain ontologies andbetween them and the core or upper ontology providing links tothe appropriate terminology for the same data items. This type of’translation’ function becomesmore compellingwhenused to viewinitial project briefs or client constraints and later when viewingthe rationale for changes as it helps all actors to understand thereasoning involved in a language they can comprehend easily. In-deed Yang and Zhang, [52] have proposed extensions to the IFCs tomap them into ontologies for the construction domain to improvethe semantic interoperability of BIMmodels in just such away. Themapping of ontology concepts into the current data model specifi-cations would be performed initially in a semi automated fashion,perhaps using tools such as those identified by Amor, [53] formap-ping between the data standards themselves, suitablymodified forthe task.

Taking the schema evolution point from Table 1, we wouldargue for a controlled process for the update and revision of bothcore and domain ontologies as suggested in [54]. This process ofcontinuous refinement of the ontologies means that at domainlevel and above, they should evolve naturallywith use by the actorsin the domains. However, ontologies would not necessarily helpwith the evolution of the design level information schema (i.e. thedrawings or models created by designers), a problemwhich wouldremain to be resolved by further research. By contrast, neither STEPnor the IFCs address these issues directly and the relatively staticrelease based versions of the standards limitwhatmight otherwisebe a route to domain level evolution to a crawl.

With respect to issues around views over model data, theExpress suite of languages employed by the STEP standards includeExpress-X, which can be used to create so called ‘views’ of Expressbased STEP models. These views are essentially newmodels basedon a new schema which has had the elements redundant for thecurrent task or context removed or otherwise transformed byaggregation, decomposition etc. Express-X is used in this scenarioto define theway inwhich the base or input schema[s] (andmodelsbased on it) is/are related to the view schema. Language constructsallow for the definition of rules about how to derive objects andproperties in the view schema from the input schema[s]. In theauthors’ opinion, Express-X adds nothing in terms of the internalsemantic expressiveness of either the base or view schemas and adedicated mapping has to be written for each set of base and viewschemas.

Where unstructured project information is concerned, the useof ontologies in tandem with other techniques drawn from in-formation retrieval/extraction could be used to automatically in-fer links between the structured and unstructured informationand indeed between items of unstructured information, basedon the links defined in the ontology. These links lend a greaterdegree of context to each item relative to the project as awhole. Benefits may also be derived from uncovering previouslyunseen linkages between various elements of project data us-ing such analysis methods. Kosovac et al., [55] and Schere &Schapke, [56] have done research work in this or closely relatedareas. Some elements of the authors’ own work have developedor used ontologies in ways similar to these. The eCognos [57,40]project for example, developed and used a construction orientedontology to augment the services that it offered as part of the


collaborative knowledge management environment also devel-oped on the project. The FUNSIEC project reviewed numerousEuropean semantic resources, compiling them into an educational‘Experience Centre’ and further conducting a feasibility study intothe production of what the project termed an ‘Open Semantic In-frastructure for the European Construction Sector (OSIECS) [58].

5. Adoption and diffusion of a construction ontology

In the following section, we discuss the reasons behind thelow adoption of IFCs, introduce the requirements for a successfulConstruction ontology, and provide an illustration of a potentialontology for the sector.

5.1. Reasons behind the low adoption of IFCs

A number of studies have been reported in the literature de-scribing various theories and models related to information tech-nology adoption, diffusion, and innovation into the workplaceand across industry. Some of these theories describe transitionprocesses and mechanisms, including Rogers’s stage model ofDiffusion of Innovations (DoI) in organizations [59]; whereas oth-ers define causality among factors to predict successful transi-tion of a technology, including Davis’s Technology AdoptionModel(TAM), [60]. TAM argues that end-user acceptance and use of infor-mation systems innovations is influenced by their beliefs regardingthe technology. In particular, it proposes that perceived usefulnessand perceived ease of use influence the use of information systemsinnovations and that this effect is mediated through behaviouralintentions to use [61]. The model highlights the critical role of ex-trinsic motivation and, in particular, expectations of task-relatedperformance gains in end-users’ adoption and use of InformationSystems innovations [61]. Roger’s Diffusion of Innovation Theoryargues that the rate of adoption of a technology is influenced bya number of factors, including: Relative advantage (the degree towhich potential adopters see an advantage for adopting the inno-vation), Compatibility (the degree to which the innovation fits inwith potential adopters’ current practices and values), Complexity(the degree of the innovation’s ease of use), Trialability (the de-gree to which potential adopters have the availability of ‘‘testing’’before adopting), and Observability (the degree to which potentialadopters are able to see observable results of an innovation).

Considering both models (TAM and DoI) in the context of IFCs,the authors discuss the various factors that may provide an initialexplanation to the low adoption of IFCs:

• Perceived usefulness of IFCs: the research community and CADsoftware industry has failed in convincing the user communityabout the usefulness in adopting IFCs. This can be attributedto (a) ineffective campaigning and awareness raising from theCAD vendors who tend to be driven by market share and com-petitiveness concerns; (b) the gap between the Construction ITresearch community and the construction end-users; (c) the na-ture of the industry which is dominated by a large proportionof SMEs that operate in a survival mode and do not have the re-sources to consider, investigate, or invest in sophisticated andcostly solutions.

• Ease of use of the IFCs (also related to Rogers’s complexity fac-tor): the user community has in its majority used CAD tomimictraditional and manual ways of producing drawings. As such,ease of use is intrinsically linked with the complexity associ-ated with the adoption of an object-oriented approach to theproduction of project documentation. It requires a paradigmshift whereby users have to adopt a lifecycle approach to dataand information integration and shift fromdocument (drawing)production tomaintaining and enriching a building information

model that serves as a basis to generate various consistent doc-umentation, including drawings. The general feeling of the in-dustry is that the adoption of the IFCs would require a steeplearning curve.

• Relative advantage: it can be argued that, based on the abovefacts, the research and CAD vendor community (as they are ul-timately the ones who are in charge of implementing the IFCs)have failed in convincing the user community about the advan-tages resulting from the adoption of the IFCs. In fact, there arelimited semantic CAD solutions, and these have been mainlyused in research and academic circles. The user communitydoes not therefore see a relative advantage in adopting the IFCs.Indeed, there is a belief that the technology is not yet matureand reliable and that there is an inherent adoption risk.

• Compatibility: this is also a major concern as the IFCs requirea paradigm shift and a migration from document centred tobuilding informationmodel oriented approaches to design. Realincompatibility concerns have been raised, as there is a lack ofpreparedness and availability of a full and complete suite of IFCcompatible ICT solutions to support the complex design stages.

• Trialability: whilst there are many commercial applicationsthat can import or export IFC format files, their native formatsare entirely different and proprietary. Thus trailability is diffi-cult as existing demonstrators are mainly academic prototypesthat can hardly be put between the hands of practitioners as(a) these are not stable and robust enough; (b) have mainlybeen developedwith open source or non-industry friendly plat-forms/solutions (e.g. Object Oriented database systems); (c) re-quire an education/training programme prior to testing so thatthe users would grasp the underlying concepts.

• Observability: A number of efforts have been made in this re-spect. For instance, the ATLAS project team in the early ninetieshave released an interesting video that illustrates the advan-tages of adopting product data technology. However, this didnot reach the whole user community and should have been ar-ticulated around a training and education initiative aimed at po-tential users from large, medium and small construction firms.Instead, this was mainly released and used amongst the re-search and academic community.

Addressing the above factors should facilitate (a) quicker uptakeof a technology, before attention from adopters fades away [62],and (b) critical mass of adoption, as there must be sufficient usersand sufficient software providers that support a standard/tech-nology before it becomes economically viable. Hence, ‘‘timing’’ and‘‘critical mass’’ emerge as key success factors.

5.2. Requirements for a successful ontology

The authors would argue that the development of an ontologyshould factor in the above considerations and, hence, be embeddedin a wider initiative aimed at raising awareness of the constructionuser community through education, training, demonstrations, etc.

Also, given the following factors: (a) the fragmented anddiscipline-oriented nature of the construction sector; (b) the var-ious interpretations that exist of common concepts by differentcommunities of practice (disciplines); (c) the plethora of seman-tic resources that exist within each discipline (none of which havereached a consensual agreement); (d) the lifecycle dimension ofa construction project with information being produced and up-dated at different stages of the design and build process with astrong information sharing requirement across organizations andlifecycle stages; a suitable ontology development methodologyshould accommodate the fact that the ontology should be specificenough to be accepted by practitionerswithin their own discipline,while providing a generic dimension that would promote commu-nication and knowledge sharing amongst these communities.


A critical analysis of the semantic resources available inconstruction, ranging from taxonomies to thesauri, combinedwithan understanding of the characteristics of the sector, have helpedformulate a set of requirements that ought to be addressed in ordertomaximize the chances of awide adoption of any ontology projectin the construction sector. These requirements are listed below:

• The ontology should not be developed from scratch but shouldmake as much use as possible of established and recognizedsemantic resources in the domain.

• The ontology should be built collaboratively in a multi-user environment: the construction sector involves severaldisciplines and communities of practice that use their ownjargon and have specialized information needs.

• There is a need to ensure total lifecycle support, as theinformation produced by one actorwithin one discipline shouldbe able to be used by others working in related disciplines.

• The ontology must be developed incrementally involving theend-users. This is important given the multi-disciplinary andmulti-project nature of the industry, and the fact that eachproject is a one-off prototype.

• The ontology should be flexible and comprehensive enough toaccommodate different business scenarios used across projectsand disciplines.

• The ontology should be user friendly, i.e., easy to use andproviding a conceptualization of the discipline/domain beingrepresented that embeds the technical jargon used in the sector.

• The ontology should be a living system and should allow forfuture expansion.

5.3. eCognos: An example of a candidate ontology for the constructionsector

Given the above requirements, an ontology was developed,referred to as eCognos [40,54]. eCognos is the acronym of anIST Framework 5 project involving the authors that developed aknowledgemanagement platformcomprising a number of servicescentred around an ontology. The latter constitutes one of the keycontribution of the project. The eCognos ontology is structured intoa set of discrete, core and discipline-oriented, sub-ontologies. Eachsub-ontology features a high cohesion between its internal con-cepts while ensuring a high degree of interoperability betweenthem. These are organized into a layered architecture (three layers)with, at a high level of abstraction, the core ontology that holds acommon conceptualization of the whole construction domain en-abled by a set of inter-related generic core concepts forming theseeds of the ontology. These generic concepts enable interoperabil-ity between specialized discipline-oriented modules defined at alower level of abstraction. Thismiddle layer of the architecture pro-vides discipline-oriented conceptualizations of the constructiondomain. Concepts from these sub-ontologies are linked with thecore concepts by generalization/specialization (commonly knownas IS-A) relationships. The third and lowest level of the architec-ture represents all semantic resources currently available, whichconstitute potential candidates for inclusion into eCognos either atthe core or discipline level.

There are a large variety of available semantic resources thatcould form the basis for building the eCognos core ontology. Theserange from classification systems to taxonomies. The latter deserveparticular attention as argued in [63]. One of the principal roles oftaxonomies is to facilitate human understanding, impart structureon an ontology, and promote tenable integration. Furthermore,properly structured taxonomies: (a) help bring substantial orderto elements of a model; (b) are particularly useful in presentinglimited views of a model for human interpretation; and, (c) play acritical role in reuse and integration tasks. Improperly structuredtaxonomies have the opposite effect, making models confusing

and difficult to reuse or reintegrate [63]. IFCs, being more recentand also the closest taxonomy currently in use in the sector,are therefore the preferred candidate semantic resource that canprovide the skeleton on which such a core ontology can be built.

A particular approach is adopted for building and/or expandingthe discipline-oriented sub-ontologies. This involves selecting andmaking use of a large documentary corpus used in the disciplineand ideally produced by the end-users. The sub-ontologies are thenexpanded and built from index terms extracted from commonlyused documents using information retrieval techniques [54]. Thiswould allow the effective capture of the jargon and its associatedsemantics usedwithin each community of practice/discipline. Thisis illustrated in Fig. 3.

An ontology of the construction domain will include conceptsdevoted to the description of the processes involved in construc-tion projects, thus the relationships between the data on the onehand, and the process within which it is used on the other becomemore explicit. It is true to say that STEP and the IFCs both definehigh level schema elements related to process and the use ofmodeldata within the process, elements which would need to be dupli-cated in an ontology. However, certainly where STEP is concerned,those elements are used to describe processes and data use withinindividual application protocols (APs) and therefore understandingthe relationships between data in the individual disciplines (whichthe APs serve) and the overall project process remains largely un-supported. The explicit links in an ontology between process anddata concepts can be used to map out a detailed context for infor-mation elements relevant to each actor’s role in the project andpresented in terms that they would normally use. The existence ofa process directly serviced by an information system all of which isontologically described allows us to create interfaces to the processfor individuals based on their role and the specific stage of the pro-cess at which they are currently working, which present relevantinformation in a timely manner again customised to the needs andlanguage of the actor.

Therefore, an ontology is not an end in itself. It should have a keyand pivotal role in enabling semantic integration across the projectlifecycle and its various disciplines. The following sections proposeinter-enterprise collaboration architectures and frameworks basedon semantic services, underpinned by ontology-based knowledgestructures.

6. Integration through ontology-based semantic services

The pragmatic ontology approach naturally leads to a service-oriented view of the world, whereby a particular discipline (andtherefore the applications that support that discipline) definesa number of services that it may offer to other disciplines,with clearly agreed semantics that can be understood by thecommunicating parties without having to change the bulk ofthe conceptualizations that they use in order to provide thoseservices. As such, ontologies should provide conceptualisationsthat can span several stages of a project while factoring in non-technical (i.e. socio-cultural and organisational) considerations.These conceptualisations can evolve and be adapted to specificcontexts that can be shared across services related to one or severaldesign and construction stages. The only changes are related to theapplications/services used within each stage.

6.1. Desktop applications versus web services

The fundamental difference between a web service and a tra-ditional desktop application (as widely available and used in theconstruction industry) is that while the former can be virtually ac-cessed ‘‘anytime, anywhere’’, the latter is only usable from withina desktop (fixed physical location) or at best through a local/widearea network using a client/server approach.Web services provide


Material

Layer Layered Element

Covering Layer

InsulationLayer

ArchitecturalLayer

StructuralLayer

LoadbearingWall

StructuralWall

Cavity Wall

Is-composedof

Is Composedof

Is Composedof

Is-composed of 1

11

1

1

1

1

StructureArchitecture

Architecture Documentation & Architectural Design related Email Archive

is-a

is-a

is-a

is-a

1..*

1..*

1..*

1..*

1..*1..* 0..*

Is-made-of

Structure Project Documentation

& Structure related Email Archive

CoreOntology

Cor

e O

ntol

ogy

Sub

-Ont

olog

yD

ocum

enta

tion

Fig. 3. Levels of abstraction of the proposed ontology.

a more flexible middleware solution that suits application inter-working, leveraging inter and intra enterprise information sys-tems. Web services are self-contained, web-enabled applicationscapable not only of performing business activities on their own,but also possessing the ability to engage other web services in or-der to complete higher-order business transactions [64]. The ben-efits of web services include the decoupling of service interfacesfrom implementation and platform considerations, the support fordynamic service binding, and an increase in cross-language andcross-platform interoperability. The challenge of this new form ofcomputing is to move from its initial ‘‘Describe, Publish, Interact’’capability to support dynamic composition of services into rein-vented assemblies, in ways that previously could not be predictedin advance [64,13]. One of the interesting features of web servicestechnology is that it provides themeans for traditional desktop ap-plications to be enhanced and upgraded to become and/or to em-ploy web services [13].

Given the shortcomings we have identified in the current prod-uct data centric approaches to integration and our suggestion thatthe use of ontologies at this level could go some way towards ad-dressing them, we would further propose that in order to havesystems that actors can interact with in a more intuitive way,ontologies have more roles to play. Here we envisage a num-ber of elementary components, each furnishing a small piece offunctionality, usually some discipline specific function, in a fullyencapsulated independent fashion. These components, publishedas Web Services could be further composed into higher-levelbusiness process components, again self-contained as per the

component based development typical of modern object orientedsystems. This model of arbitrary combinations of process compo-nents or e-processes as one might call them, allows for greaterflexibility in the definition and production of business systems tosupport construction projects. Ontologies play their role not onlyat the basic level of a semantic integration layer over the data asdetailed previously, but also (a) as a means to describe the con-cepts and relationships inherent in the processes of constructionprojects and (b) as a means to articulate at a semantic level theprecise nature of an offered service.Working at this higher processoriented level, we begin to see opportunities for resolving some ofthe lifecycle and context shortcomings of current data models.

The OSMOS [65], C-Sand [61] and eCognos [40,5,57] projects inwhich the authors were involved all employed architectures fea-turing multiple interoperating services to furnish their functional-ity to varying degrees. The success of these projects demonstratesthe utility of the orchestrated service approach whilst eCognos, aspreviously mentioned, also featured an ontology to augment itsservices with semantic capabilities. The systems developed underthe eCognos and C-Sand projects also featured the ability to con-sume arbitrary web services for use ‘on the fly’. The developmentof this feature did not however extend to any automated notion ofwhat those arbitrary services ‘were’ or ‘did’, which rather limitedits usefulness. Thus we believe that extending the work to encom-pass services semantically described bymeans of ontologieswouldallow for a more automated integration and orchestration to takeplace, particularly when ontologies are also used to map the ser-vices to the business process being served.


6.2. Support for semantic e-Processes

The means by which to aggregate and implement a num-ber of Web Services into a larger business process orientedservice are provided through the Business Process Execution Lan-guage (BPEL [66]). The BPEL specification defines constructs simi-lar to a simple programming language such as loops, assignments,branches etc, with which to define the flow of calls between acollection of orchestrated services involved in a modeled process.Processes may be graphically modeled using one of several mod-eling standards such as BPMN (Business Process Modeling No-tation) [67]. BPMN is the natural complement to BPEL for theinitial design of the aggregated and orchestrated services with sev-eral translators being available to turn a BPMN model into exe-cutable BPEL code.

The whole Web Services stack comprises several other stan-dards including those for initial technical description of services(WSDL – Web Services Description Language) [68], standardsfor securing communications between services and clients (WS-Security), and standards for transaction demarcation and manage-ment (WS-Transaction). One important protocol for the future ofWeb Services is the Universal Description, Discovery and Integra-tion protocol (UDDI [73]), which allows for the publication and dis-covery of services described using WSDL on the Web. UDDI has aproblem in that whilst it is possible to publish a service descrip-tion and have it searchable by others, it does not make explicitwhat the service is for in language that a machine can understand.Thus it is that a human must currently decide whether a partic-ular service is suitable for their business’ needs by manually ex-amining both the technical and textual descriptions of a servicefor compatibility. Removing this manual intervention would allowUDDI to bemuchmore useful than it is today andwould permit thetype of process oriented services we envisage to be assembled in amore automated fashion. It is here that much current research anddevelopment work is concentrated under the Semantic Web Ser-vices banner. Standards to describe what a service is for and whatthe various inputs and outputs actually mean in machine inter-pretable form are being developed with OWL based ontology forWeb Services (OWL-S [69]) and the Web Services Modeling On-tology (WSMO [70]). Both of these standards define ontologicalconstructs for describing services and allow external ontologies tobe used in the description of service parameters. It is in this rolethat domain specific ontological concepts such as those defined bythe FUNSIEC project [58] or eCognos [40,54] can be employed todescribe services for particular business fields. Together with thelower level Web Services protocols, these ontologies allow for thesemi automated composition of aggregate servicesmodeled in linewith the business process requirements of specific domains.

While web service technology presents some interesting andpromising features, there are a number of issues hindering thewide adoption of this technology, including data quality assurance,quality of service (continuity and recovery plans), as well as a trust,authentication, security, validation, and certification framework(authentication and trust). Addressing these limitations wouldconfer web service technology the industrial robustness thatwould promote its wide adoption on projects. The next sectiondiscusses the dynamic nature of constructionproject processes andthe requirements this places on the proposed e-process enabledcomputing environment.

7. Lifecycle dimension and support for the dynamic and long-lasting nature of e-Processes

The design stage of a project involves interesting examples oflong-lasting processes. This section (a) discusses the limitations

of service-based process approaches in addressing the complex-ity of architectural design, characterised by long-running coop-erative processes, (b) provides a comprehensive framework thatsummarises the above shortcomings, and provides a potential e-Platform solution for the construction industry, and (c) suggests abusiness model based on the ‘‘application service provider (ASP)model’’ for the deployment of the e-Platform solution for the con-struction industry.

7.1. Design process characteristics

The multiplicity of circumstances governing the decisionmaking processes inherent in architectural design leave scope fornumerous misunderstandings, unforeseen difficulties created byinappropriate or ill-conceived information, changes and decisionswhich fail to propagate amongst all interested parties. Further,these circumstances are commonly compressed into short time-frames featuring periods of intense activity in which many deci-sions are made. Moreover, design is a predictive activity, that hasto be planned and instrumented, and for which actions that will beimplemented are defined beforehand. At the same time, design isa reactive activity that evolves and adapts as its content changeswith the environment and with the personality of the actors thatconduct it. All the complexity of the design therefore lies in this du-ality. It is therefore agreed upon that if design steering consists oforganizing and planning tasks with already identifiedmechanismsand results; it also consists in managing events, actions and situ-ations that are not initially known and formalized. The success orfailure of a project is often explained by themanner in which thesedifferent unplanned situations are managed and controlled.

The design process is currently supported by a number of soft-ware applications, including CAD and related engineering soft-ware. It can be modeled as a dynamic, long-lasting, process.However, there exist several limitations of service process ap-proaches that hinder the effective adoption of such a paradigm.Long running cooperative processes are subject to evolutions andchanges of differing nature: processmodel evolution due to changein the environment (change in the law, change in the method-ology), process instance evolution (or ad-hoc evolution) due tospecific events occurring during a given process execution (de-lay, newly available or missing resources) or partnership evolu-tion at execution time having an impact on part of the process.These shortcomings require essential advances and improvements,including:

• Tracking of history of changes: change management is animportant issue in long lasting processes. When a processmodel (or an abstract process) is changed, it may be importanttomigrate running processes to reflect these changes. However,this migration is sometimes only feasible under certainconditions, and must be implemented dynamically.

• Partner change during process execution (dynamic change ofpartner, with partial fulfilment of choreography): during a longlasting process, a partner may fail to complete a conversation,or even disappear. In this case, a new partner has to beselected, dynamic re-composition has to occur and part of theexecution may have to be restarted. It is essential that changeof partners be dynamically supported in the context of theexecuted choreography.

• Partial rollback (check pointing): events such as dynamic cha-nge of partners may require a process to be partially rolled backand re-executed with a new partner. This partner may evenbenefit from the previous partial execution. Partial rollback orcompensation may also be triggered by a change in the pro-cess. These scenarios are unsupported or ill-supported by thecurrent BPEL specification. Partial rollbackwill require adapting


Fig. 4. An ontology-based framework for e-Process execution and management.

the Business Activity transactionmodel to allow amore flexibleapproach than the simple open nested transactions.

• Process evolution (unpredictable event management, dynamicprocess evolution): during a long lasting process, events mayoccur such as unexpected delay or resources evolutions thatrequire more or less important changes in the process. Thesechanges have to be done while ensuring the correctness of theprocess itself. The kind of changes that have to be tackled con-cern adding or removing operations in the process, change inthe ordering of the steps, changes in the relationships with thepartners (policy evolution). Somework regarding dynamic pro-cess evolution has been done already in the area of workflowmanagement systems, which would be of benefit if adapted toBPEL processes. Ad hoc changes are required to ensure the reli-ability and the validity of the resulting process.

BPEL as it is defined does not support these kinds of processmodel evolution and even less ad-hoc evolutions. This is a realproblem for long running processes as experienced during the de-sign stage of a project where external and internal unexpectedevents may require adaptation and evolution. It is worth notingthe pivotal role of a construction ontology in resolving many ofthe above limitations, in particular those related to semantic com-patibility between services (as illustrated in Section 7). Since in-dividual web services are created in isolation, their vocabulariesare often rife with problems having abbreviations, different for-mats, or typographical errors. Furthermore, two terms with differ-ent spellings may have the same semantic meaning, and thus areinter-changeable. Diversematching schemes have been developedto address these semantic resolution problems, including in thecontext of the eContent FUNSIEC project [58]. In general, match-ing approaches may fall into three categories: (a) Exact match us-ing syntactic equivalence; (b) Approximate match using distancefunctions (TF-IDF, Jaccard, SoftTF-IDF, Jaro, or Levenstein distance);and (c) Semantic match using ontologies. The latter is the authors’preferred approach as it provides the possibility to reason aboutweb services, and to automate web services tasks, like discoveryand composition [15].

7.2. Proposed e-Platform solution

Fig. 4 illustrates a comprehensive framework that summarisesthe above shortcomings and issues identified in the paper, andprovides a potential e-Platform solution for the constructionindustry. This has been inspired from the authors’ research. Theunderlying web-service infrastructure has already been developedas reported in [10], while some of the suggested services havealready been specified and prototyped, including the Ontologyservice [61] and the Semantic compatibility service [15]. A briefexplanation of each service is given below:

• Semantic compatibility service: this, as described earlier, deter-mines whether two services can inter-work prior to invocation.In essence, the semantics of their underlying data structures arelooked up in an ontology and semantic relatedness checks areperformed.

• Process monitoring service: this provides a means to monitorthe various states a process goes through and anticipate poten-tial problems, including availability of services.

• Ontology service: this provides the functionality required tomake the selected ontology available to the other services,which may require it. It provides an interface to query the on-tology, including its concepts, and semantic relationships.

• Intelligent resource discovery service: This service makes useof a categorisation sub-service that provides a context andcriteria-based categorisation of the information held within theUDDI registry for effective use and mining by potential busi-ness partners. It provides a standardised interface to query andmakes use of the Intelligent Categorised UDDI Registry gener-ated through the categorisation sub-service.

• Composition service: this provides the interface that enablesthemodeling of an e-Process. This is achieved through the com-position of discrete services.

• Process execution service: this is used to execute a modeled e-Process through the composition service, and provides an inter-face to control and interfere with its execution following prob-lems identified by the process monitoring service.


Ongoing Long

Fig. 5. A proposed preliminary roadmap.

7.3. A proposed business model

We would also suggest a business model based on the appli-cation service provider model for the deployment of the technicalsolution. Here we envisage three roles as described in [4]:• Service provider — any organisation having services (specif-

ically web services) that they wish to monetise and offer tothird parties for their consumption. Providers register their ser-vices with the aggregator for publishing and composition intoe-Processes.

• Service aggregator/host — An organisation with responsibilityfor hosting the infrastructure defined in the middle layer ofFig. 5. The aggregator composes business systems (e-Processes)from the offerings of registered service providers tailored tothe requirements of particular projects or organisations (real orvirtual) and their business processes.

• Service client — any organisation requiring business systemfunctionality to support their business processes.

It is believed that this model will aid construction sector SMEsin adopting technology which may otherwise be out of their reacheither technically or financially. We do not however prescribe whomay take on the individual roles and indeed envisage that single or-ganisations, particularly large technologically sophisticated ones,might encompass elements of all of them. Centralising infrastruc-ture in this way has the additional benefit of providing a singlepoint of contact for service, support and legal/contractual issuesfrom the point of view of service clients.

8. A preliminary roadmap for the purpose of semantic service-based e-Construction

The construction industry is different from other large indus-tries since it relies on a very high proportion of SMEs involved inboth off-site management, design and procurement and on-sitefabrication services. Such SME companies tend only to be presentin a project during their part of the activity and this discontinuityof involvement is a particular challenge for the industry in relationto the adoption of the right business models and modes of projectoperation. Certainly the industrialisation of construction cannotbe done in exactly the same way as in other sectors. This sectionproposes an initial roadmap for the adoption of the proposed ap-proach. Key features of the industry are provided, followed by aproposal for an alliance model. An initial staged roadmap is thenformulated.

8.1. A proposed alliance model for the industry

The general picture of construction is therefore of an industrythat is a pyramid with control being in the hands of large playerswith a large base of SMEs relatively weak in influencing theearly important decisions in projects because they are usually notappointed and in place. The expertise of SMEs cannot thereforebe brought to bear at the conceptual and feasibility stages inconstruction. Also, some industry major players are reducing thecircle of specialists and sub-contractors they use. This has animpact on the supply/value chain by forcing the emergence oftransient knowledge based alliances. However, these are highlylikely to be dominated by large industry players, where SMEs haveno influential role. Hence, SMEs should anticipate this trend byforming cooperative transient production networks [71].

The way the authors view the future is for SME alliances tobe established, forming knowledge infused virtual networked or-ganisations that can ‘‘punch at a higher weight’’ than the individ-ual SMEs. The alliance brings together SMEs in a relationship oftrust to provide holistic competence to a field of activity — for in-stance, ‘‘energy consumption, energy performancemonitoring andintegration of energy resources’’. The alliance would market itselfwith this capability and ultimately achieve a recognised industry‘‘branding’’ much as a ‘‘big name’’ player. Moreover, because thealliance SMEs would work frequently together (though not exclu-sively), they can (a) standardise processes and information flows,(b) improve sharing of information and re-use of best practice, (c)create value out of the knowledge possessed across the alliance,thus making investment in collaborative technologies highly at-tractive and productive. A key function in an alliancewill be to bro-ker, manage and integrate the alliance acting as the conduit withclients andmanaging the quality of solutions achieved. The authorsview an SME alliance as a kind of Virtual Factory, with different al-liance members contributing their often much specialised knowl-edge on particular aspects to virtually configure a ‘‘best’’ solutionagainst requirements and regulations.

Thus, the authors argue a transformation of production businessmodels for Construction SMEs, to achieve economies of scale inthe production of standardised processes and product approacheswith economies of scope in various stages of assembly in order toprovide flexibility to satisfy customer choices [72].

This is where service-oriented architectures have the abilityto provide the underpinning middleware infrastructure that cansupport the operations of virtual alliances. Ontologies will have a


pivotal role in this, as they will provide the semantic backbone toenable concept integration between actors and applications.

8.2. An initial adoption roadmap

E-processes are typically designed, developed, and deployedby enterprises that want to compose internal capabilities withthird-party capabilities, either for internal use or to expose themas (complex, value-added) e-services to customers. For both e-processes and e-services, the need for companies to exposeinternal details of how they run their business is not envisioned.Still, companies should be able to express, in a standard format,the interaction aspects for their service offer as well as for theirservice needs. A common language is crucial for the assessmentof the compatibility between the interaction processes offered byan e-service provider and those expected by the designer of ane-process. The effectiveness and efficiency of business processesimpact directly the profitability of a company. It is in the bestinterest of e-service providers as well as e-service consumers tounderstand the operational requirements for their cooperation.This can best be achieved in the context of an alliance. However,as traditional processes are designed around the operationalmodel of customised business applications, e-processes should bedesigned around e-services. A clear understanding of the businessinteraction model of an e-service is paramount, and this shouldbe facilitated by the service provider. Migrating to e-Processesinvolves the three following stages illustrated in Fig. 5 and brieflydescribed below:

• Phase 1: integration of existing internal assets. Enterprises(including SMEs) that form an alliance should align theirprocesses to the business needs and vision of the allianceand work towards reducing costs and improving businessprocess execution quality and speed (operational excellence).Process automation and management technologies enablethe separation of business, resource and application logics.Processes can be controlled, managed, and evolved separatelyfrom the applications. In this phase, resources remain internalto the enterprise.

• Phase 2: static integration with partner processes on a case-by-case basis. By incorporating e-services provided by businesspartners into an e-process (within or across alliances), anenterprise can create processes that utilize external resourcesoffered by other alliance members. However, in this phase,service selection and invocation is still performed in an ad-hoc way, and requires preliminary agreements (from business,legal, and technical perspectives) between the cooperatingalliance companies.

• Phase 3: dynamic integrationwithnegotiation,with companies.Beyond such static use of external services, fully dynamic e-processes make decisions each time they are executed in orderto invoke the best available service that can fulfil the customer’sneeds. The traditional design-deploy cycle of phases 1 and 2 ischanged to e-Processes that cannot be anticipated in advance.This will make full use of the approach proposed in the paper,centred around the use of ontology.

However, for the above roadmap to work, the constructionIT and software (including CAD) vendor communities shouldplay a key role in delivering a robust base technology (serviceinfrastructure), supported by key adopters (including engineeringdesign consultants and contractors), and work in unison towardsaddressing the identified factors that have affected the wideadoption of previous initiatives, including: relative advantage (thedegree to which potential adopters see an advantage for adoptingthe innovation), compatibility (the degree to which the innovationfits in with potential adopters’ current practices and values),

complexity (the degree of the innovation’s ease of use), Trialability(the degree to which potential adopters have the availabilityof ‘‘testing’’ before adopting), and Observability (the degree towhich potential adopters are able to see observable results of aninnovation). It is worth re-emphasising the importance of ‘‘timing’’and ‘‘critical mass’’ in ensuring the successful deployment andadoption of this technology, as argued earlier in the paper.

9. Conclusion

It has been argued that, for successful integration in construc-tion through IT, attention needs to be paid to supporting processesthrough service oriented approaches; and to improving the humancommunication aspects and migrating existing information sys-tems through work on ontologies. Rather than attempting to cre-ate a vision of common data standards that need to be achievedin whole for benefits to be seen, these approaches provide thepotential to realize incremental benefits for the industry throughprogressive automation of processes,whichmay bemore palatableto the industry. In this scenario, existing work on product mod-els is no longer directed towards the exchange of complete datasets between applications, but provides the cornerstone for defin-ing services that fit into a service-oriented architecture to supportconstruction processes; for defining ontologies that can help to in-tegrate and migrate valuable, existing, unstructured informationand knowledge; and for focussing directly on the interactions be-tween different human actors and disciplines in the constructionindustry.

A technological solution, it has been shown, has to demon-strate capability of supporting the central project (includingdesign) business processes, allow integration of systems and in-teroperability between disparate applications and enable theman-agement of interactions between individuals and teams, whilst atthe same time taking into account the fact that the industry isdominated by SMEs and operates within tight financial margins.The proposed approach will essentially provide a scalable and userfriendly environment to support teamwork in the sector by: (a)delivering to clients customised solutions in the form of web ser-vices maintained by a dedicated application service provider; (b)providing an alternative to the traditional licensingmodel for soft-ware provision by introducing a model based on service rental oroffered on a pay-per-use basis; (c) providing a change of focusfrom ‘‘point to point’’ application integration to service collabora-tion and inter-working; (d) delivering higher-order functionality,composed from elementary services, providing direct support forbusiness processes; (e) providing a ubiquitous dimension to busi-ness processes, as services can be invoked anytime, anywhere froma simpleweb-browser; (f) enabling a single point of contact for ser-vice and client support. Furthermore some of the organisationaland technical limitations which constrain the current implemen-tation and use of web services, such as trust, security, quality as-surance, online fee transaction, and contingency handlingwill haveto be alleviated in order to realise a network of industrial strengthservices.

The paper argues that ontologies provide a richer conceptuali-sation of a complex domain such as construction compared to ex-isting product data standards. An ontology should be viewed as aliving system. The issue of the existence of a unique ontology foran entire sector remains open. This suggests thatwhile the eCognosCore Ontology forms a robust basis for interoperability across thediscipline-oriented ontologies, the latter will need adaptation andrefining when deployed into an organization and used on projects.Another issue that can be raised is that related to the adoption ofuser specific views or perspectives on the global ontology. In fact,in many instances, some actors might be required as part of theirjob to deal with more than one discipline ontology to conduct a


task. This necessitates some flexible mechanisms that can enablethe rapid combination of two or more discipline ontologies intoa single view/perspective. It is hoped that the paper will stimu-late thinking and discussion about the evolution of data productsto knowledge-rich ontologies, and their use in the context of con-struction projects to support seamless eProcesses.

It is envisaged that the technology solutions (i.e. ontologyand service architecture) as well as the adoption and diffusionissues discussed in the paper can be furthered and extended toother industry applications and sectors. Also, the authors wouldencourage the construction research community to detail anddevelop further the proposed (initial) roadmap, for instance byconsidering the Capability andMaturityModel (CMM)widely usedin the software industry. These form the research efforts in whichthe authors are currently engaged, and will be reported in relatedpublications.

Acknowledgements

The authors would like to acknowledge the financial supportfrom the European Commission under the IST and eContentprograms, aswell as the EPSRC on the ongoing EP/E001882/1 grant.

References

[1] Rezgui Y. Review of information and knowledge management practices stateof the art in the construction industry. Knowledge Engineering Review 2001;16(2).

[2] ISO 10303-1:1994 Industrial automation systems and integration – Productdata representation and exchange – Part 1: Overview and fundamentalprinciples. International Standards Organization 1994. TC 184/SC 4. Availableon-line at: http://www.iso.ch/cate/d20579.html.

[3] IAI. International alliance for interoperability. IAIWeb Site, web page at http://www.iai-international.org [accessed 15.07.2007].

[4] Boddy S, Rezgui Y, Cooper GS,Wetherill M. Computer integrated construction:A review and proposals for future direction. Advances in Engineering Software2007;38(10):677–87.

[5] Rezgui Y, Zarli A. Paving the way to digital construction: A strategicroadmap. Journal of ConstructionManagement andEngineering (Journal of theAmerican Society of Civil Engineering) 2006;132(7):767–76.

[6] Rezgui Y, Cooper G, Brandon P. Information management in a collaborativemultiactor environment: The COMMIT approach. Computing in Civil Engineer-ing 1998;12(3):136–44.

[7] Cooper G, Cerulli C, Lawson BR, Peng C, Rezgui Y. Tracking decision-makingduring architectural design. ITcon 2005;10:125–39. http://www.itcon.org/2005/10.

[8] Meziane F, Rezgui Y. Document management methods based on similaritycontents. Information Sciences 2004;158:15–36.

[9] Rezgui Y. Ontology driven knowledge management using informationretrieval techniques. Computing in Civil Engineering 2006;20(3).

[10] Rezgui Y. Exploring virtual team-working effectiveness in the constructionsector. Interacting with Computers 2007;19(1):96–112.

[11] Rezgui Y. (b) Role-based service-oriented implementation of a virtualenterprise: A case study in the construction sector. Computers in Industry2007;58(1):74–86.

[12] Rezgui Y. (c) Knowledge systems and value creation: An action researchinvestigation. Industrial Management and Data Systems 2007;107(2):166–82.

[13] Rezgui Y, Medjdoub B. A service infrastructure to support ubiquitousengineering practices. In: The 8th IFIP working conference on virtualenterprises. 2007.

[14] Rezgui Y, Meziane F. A web services implementation of a user centred knowl-edge management platform. Journal of Intelligent Information Technologies2005;1(4).

[15] Rezgui Y, Nefti-Meziani S. Ontology-based dynamic composition of servicesusing semantic relatedness and categorisation techniques. In: ICEIS: 9thinternational conference on enterprise information systems. 2007.

[16] Ellis JHM, Kiely JA. Action Inquiry Strategies: Taking stock andmoving forward.Journal of Applied Management Studies 2000;9(1):83–94.

[17] Dubois AM, Flynn J, Verhoef MHG, Augenbroe F. Conceptual modelingapproaches in the COMBINE project. Presented in the COMBINE final meeting1995. http://erg.ucd.ie/combine/papers.html.

[18] Eastman CM. Modeling of buildings: Evolution and concepts. Automation inConstruction 1992;1:99–109.

[19] Brown A, Rezgui Y, Cooper G, Yip J, Brandon P. Promoting computer integratedconstruction through the use of distribution technology. ITcon 1996;1:51–67.

[20] Autodesk. Building information modeling. 2007. Available on-line at http://www.autodesk.com/buildinginformation/.

[21] Daniell J, Director SW.Anobject oriented approach to CAD tool controlwithin adesign framework. In: Proceedings of the 26th ACM/IEEE conference on designautomation. 1989.

[22] Eastman C, Wang F, You S-J, Yang D. Deployment of an AEC industry sectorproduct model. Computer-Aided Design 2004;37(12):1214–28.

[23] Bosch KO, Bingley P, van der Wolf P. Design flow management in the NELSISCAD framework. In: Proceedings of the 28th conference on ACM/IEEE designautomation. 1991.

[24] Eastman CM. Information exchange architectures for building models.In: Lacasse, MA, Vanier, DJ, editors. Durability of building materials andcomponents. vol. 8. Ottawa (Canada): Institute for Research in Construction.p. 2139–56.

[25] Rezgui Y, Cooper G. A proposed open infrastructure for construction projectdocument sharing. ITcon 1998;3:11–25.

[26] Gu P, Chan K. Product modeling using STEP. Computer-Aided Design 1995;27(3):163–79.

[27] Fitzgerald B, Howcroft D. Competing dichotomies in IS research and possiblestrategies for resolution. In: Proceedings of the international Conference oninformation systems. 1998. p. 155–64.

[28] Mannisto T, Peltonen H, Martio A, Sulonen R. Modeling generic productstructures in STEP. Computer-Aided Design 1998;30(14):1111–8.

[29] GielinghWF. General AEC referencemodel. Tech. rep. P.O. Box 46 2600 AA (theNetherlands): BI-88-150; October 1988.

[30] Turner J. AEC building systems model. Working paper ISO/TC/184/SC4/WG1;October 1988.

[31] Bohms M, Tolman F, Storer G. ATLAS, A STEP towards computer integratedlarge scale engineering. Revue Internationale de CFAO 1994;9(3):325–37.Available on-line at: http://www-uk.research.ec.org/esp-syn/text/7280.html.

[32] Björk B-C. RATAS project — Developing an infrastructure for computer-integrated construction. Journal of Computing in Civil Engineering 1994;8(4):400–19.

[33] Froese T, Paulson B. OPIS, an object-model-based project information system.Microcomputers in Civil Engineering 1994;9:13–28.

[34] Björk B-C, Penttilä H. A scenario for the development and implementation ofa building product model standard. Advances in Engineering Software 1998;11(4):176–87.

[35] Björk B-C. Basic structure of a proposed building product model. Computer-Aided Design 1998;21(2):71–8.

[36] Giannini F, Monti M, Biondi D, Bonfatti F, Daniela Monari P. A modeling toolfor the management of product data in a co-design environment. Computer-Aided Design 2002;34(14):1063–73.

[37] VTT. IFC model server http://ve.cic.vtt.fi/IMSvr/index.html.[38] Fischer M, Hartmann T, Rank E, Neuberg F, Schreyer M, Liston K, et al.

Combining different project modeling approaches for effective support ofmulti-disciplinary engineering tasks. In: Brandon P, Li H, Shaffii N, Shen Q,editors. INCITE 2004 — International conference on information technologyin design and construction. 2004. p. 167–82.

[39] KamC, FischerM,Hänninen R, KarjalainenA, Laitinen J. The productmodel andfourth dimension project. ITcon 2003;8:137–66 [special issue IFC - Productmodels for the AEC arena].

[40] Lima C, El-Diraby T, Stephens J. Ontology-based optimisation of knowledgemanagement in e-construction. ITcon 2005;10:305–27.

[41] NIBS. National Institute of Building Sciences. BIM CommitteeWeb site at http://www.nibs.org/BIMcommittee.html [accessed 13.07.2007].

[42] Corcho O, Fernando-Lopez M, Gomez-Perez A. Methodologies, tools andlanguages for building ontologies. Where is their meeting point? Data andKnowledge Engineering 2003;46:41–64.

[43] Gruber T. Towards principles for the design of ontologies used for knowledgesharing. International Journal of Human Computer Studies 1994;43(5–6):907–28.

[44] Studer R, Benjamins V, Fensel D. Knowledge engineering: Principles andmethods. IEEE Transactions on Data and Knowledge Engineering 1998;25:161–97.

[45] Wenger E, Snyder W. Communities of practice: The organizational frontier.Harvard Business Review 2000;(Jan.–Feb.):139–45.

[46] Checkland P, Scholes J. Soft systems methodology in action. New York: JohnWiley & Sons, Inc.; 2000.

[47] Bazjanac V. Building energy performance simulation as part of interoperablesoftware environments. Building and Environment 2004;39(8):879–83.

[48] Estman CM. Building product models. Boca Raton (FL, USA): CRC Press; 1999.[49] Rezgui Y, Brown A, Cooper G, Aouad G, Kirkham J, Brandon P. An integrated

framework for evolving construction models. The International Journal ofConstruction Information Technology 1996;4(1):47–60.

[50] Hassanien Serror M, Inoue J, Adachi Y, Fujino Y. Shared computer-aidedstructural design model for construction industry (infrastructure). Computer-Aided Design 2008;40(7):778–88.

[51] Kim J, Pratt MJ, Iyer RG, Sriram RD. Standardized data exchange of CADmodelswith design intent. Computer-Aided Design 2008;40(7):760–77.

[52] Yang QZ, Zhang Y. Semantic interoperability in building design: Methods andtools. Computer-Aided Design 2006;38:1099–112.

[53] Amor RW. Supporting standard data model mappings. In: Proceedings of EC-PPM 2004. p. 35–40.

[54] Rezgui Y. Text based domain ontology building using tf-idf andmetric clusterstechniques. Knowledge Engineering Review 2007;22(4):379–403.

http://www.iso.ch/cate/d20579.html







http://www.iai-international.org




http://www.itcon.org/2005/10






http://erg.ucd.ie/combine/papers.html







http://www.autodesk.com/buildinginformation/





http://www-uk.research.ec.org/esp-syn/text/7280.html









http://ve.cic.vtt.fi/IMSvr/index.html








http://www.nibs.org/BIMcommittee.html









[55] Kosovac B, Froese TM, Vanier DJ. Integrating heterogeneous data representa-tions in model-based AEC/FM systems. In: Gudnason G, editor. Constructioninformation technology 2000, Proceedings of CIT 2000 - The CIB-W78, IABSE,EG-SEA-AI international conference on construction information technology.2000, vol. 2. Icelandic Building Research Institute; 2000. p. 556–67.

[56] Scherer RJ, Schapke S-E. Constructing building information networks fromproprietary documents andproductmodel data. In: Scherer RJ, Katranuschkov,P, Schapke, S-E, editors. Proceedings of cib-w78 2005 22nd conference oninformation technology in construction. p. 343–8.

[57] Wetherill M, Rezgui Y, Lima C, Zarli A. Knowledge management for theconstruction industry: The eCognos project. ITcon 2002;(7):183–96 [specialissue ICT for Knowledge Management in Construction].

[58] Barresi S, Rezgui Y, Lima C, Meziane F. Architecture to support semantic re-sources interoperability in interoperability of heterogeneous information sys-tems. In: Proceedings of the first international workshop on Interoperabilityof heterogeneous information systems. 2005. p. 79–82.

[59] Rogers EM. Diffusion of innovations. 4th ed. New York: Free Press; 1995.[60] Davis FD. User acceptance of information technology: System character-

istics, user perceptions and behavioural impacts. International Journal ofMan–Machine Studies 1993;18(3):475–87.

[61] Wetherill M, Rezgui Y, Boddy S, Cooper GS. Intra- and inter-organizationalknowledge services to promote informed sustainability practices. Journal ofComputing. in Civil Engineering 2007;21(2):78–89.

[62] Moore GA. Crossing the chasm: Marketing and selling technology products tomainstream customers. New York: Harper Business; 1991. p. 223.

[63] Welty C, Guarino N. Supporting ontological analysis of taxonomic relation-ships. Data and Knowledge Engineering 2001;39(1):51–74.

[64] Ferris C, Farrell J. What are web services. Communications of the ACM 2003;46(6):31.

[65] Rezgui Y, Zarli A, Bourdeau M, Cooper G. Inter-enterprise informationmanagement in dynamic virtual environments: The OSMOS approach. In:Proceedings of the CIT2000 – The CIB-W78, IABSE, EG-SEA-AI internationalconference on construction information technology. 2000. p. 731–41.

[66] OASIS. Web service business process execution language technical introduc-tion. Available at http://www.oasis-open.org/committees/download.php/23068/WS-BPEL%20Technical%20Overview%20for%20Developers%20and%20Architects%20-%20Part%201%20%28Frank%20Ryan%29.pdf [accessed17.07.2007].

[67] OMG. Object management group/business process management initiative.OMG website at: http://www.bpmn.org/ [accessed 01.06.09].

[68] W3C. Web services description language. W3C Note, 15th March 2001. W3CWebsite at http://www.w3.org/TR/wsdl [accessed 01.06.2009].

[69] OWL-S Coalition. OWL-S Coalition web site. Web page at http://www.daml.org/services/owl-s/1.0/ [accessed 16.07.2007].

[70] W3C. Web services modeling ontology. W3C Member Submission. W3C Website at http://www.w3.org/Submission/WSMO/ [accessed 16.07.2007].

[71] Hagel J, Armstrong AG. Net gain — Expanding markets through virtualcommunities. Boston (Massachusetts): Harvard Business School Press; 1997.

[72] Rezgui Y, Miles JC. Transforming SME strategies via innovative transientknowledge-based alliances in the construction sector. In: Proc. 7th IEEEinternational conference on industrial informatics. 2009.

[73] OASIS. OASIS UDDI V3.0 specification.Web page at http://uddi.org/pubs/uddi_v3.htm [accessed 17.07.2007].

http://www.oasis-open.org/committees/download.php/23068/WS-BPEL%20Technical%20Overview%20for%20Developers%20and%20Architects%20-%20Part%201%20%28Frank%20Ryan%29.pdf












http://www.bpmn.org/




http://www.w3.org/TR/wsdl






http://www.daml.org/services/owl-s/1.0/








http://www.w3.org/Submission/WSMO/






http://uddi.org/pubs/uddi_v3.htm







past, present and future of information and knowledge sharing in the construction industry: towards...

Documents