A study of ontology-based risk management framework of construction projects
through project life cycle
H. Ping Tserng
a,⁎
, Samuel Y.L. Yin
a,b, R.J. Dzeng
c, B. Wou
d, M.D. Tsai
e, W.Y. Chen
a aDepartment of Civil Engineering, National Taiwan University, No. 1 Roosevelt Road, Sec. 4, Taipei, Taiwan
bRuentex Group, 14F., No. 308, Section 2, Bade Road, Taipei, 104, Taiwan c
Department of Civil Engineering, National Chiao-Tung University, 1001 Ta-Hsieu Road, Hsinchu, 30050, Taiwan
d
Department of Civil Engineering, National Chiao-Tung University, 1001 Ta-Hsieu Road, Hsinchu,30050, Taiwan
e
Division of Construction Engineering and Management, Department of Civil Engineering, National Taiwan University, No. 1 Roosevelt Road, Section 4, Taipei, Taiwan
a b s t r a c t
a r t i c l e i n f o
Article history: Accepted 13 May 2009 Keywords:
Process-oriented knowledge management Knowledgeflow
Project risk management Workflow
Engineering decisions Ontology
The process knowledge assets make a substantial contribution to the risk management (RM) for contractors in the construction phase. To effectively reuse these assets, knowledge extraction becomes a significant research area. This paper was designed to explore an approach to conduct knowledge extraction by establishing project risk ontology. Specifically, the study proposed the ontology-based risk management (ORM) framework to enhance the RM performance by improving the RM workflow and knowledge reuse. The ORM framework facilitated the identification, analysis, and response of project risks. This study validated the ORM framework through a case demonstration. Through the implementation and application, the results demonstrated that the ORM framework was able to apply to the RM workflow for contractors, and more importantly, it greatly increased the effectiveness of project RM.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
The characteristics of the construction industry include product uniqueness, on-site production, and ad hoc project teams with high
turnover rate [34,52]. Subsequently, it has been difficult for the
construction industry to coordinate, store, and reuse knowledge that is obtained between the organization and its individuals. Therefore, the construction industry needs to acquire, store and reuse knowledge in order to increase project performance. Previous studies had suggested the organization should conduct knowledge management
through methods of project reports or lessons learned[15,29,42].
In addition to the aforementioned approach, Process-Oriented Knowledge Management (POKM) emphasizes that, through the
com-bination of knowledge management and workflow process, the
organization could increase project performance and purposely
accu-mulate knowledge for future usage [40,24]. Recently, construction
studies have explored knowledge management from the process-oriented perspective and studied the effectiveness of the POKM
application in: safety management[5,12], design management[4,30],
and facility management[44]. The results revealed that these studies
primarily applied information technology (IT) tools to integrate the
workflow and knowledge management. Knowledge management was
mainly conducted to capture explicit and tacit knowledge related to the
workflow to help the user acquire and reuse knowledge with the
standard operation procedure within the workflow.
In a previous research paper by the writers[54], the case study
verified the POKM model could apply to the contractor's risk
management (RM) workflow and therefore enhance the RM
perfor-mance. Simultaneously, the project risk knowledge base could be built. Following the building of the knowledge base, an inquiry into how to extract and reuse the knowledge base effectively emerged as another important research issue. This study proposed ontology as the solution to investigate this issue. More importantly, the study would concentrate in the project risk knowledge base to study and verify the knowledge extraction model effectively. The study proposed an approach combining expert interview and information retrieval (IR) algorithms to extract the knowledge and develop the project risk ontology. Subsequently, the ontology-based risk management (ORM) framework is developed.
Based on literature review, the construction industry's project environment was usually exposed to a higher degree of risk and faced a
significant amount of uncertainties[2,7]. Under such conditions, those
decisions made by engineers and project managers were generally
under uncertainty [3]. Consequently, project performance for the
construction project was subject to risk factors and most projects failed
to deal with the risk[18,22]. In particular, during the construction
phase of project life cycle, the contractor not only faced risks produced from limited experiences in construction and project execution itself,
⁎ Corresponding author. Tel.: +886 2 23644154; fax: +886 2 23661640. E-mail addresses:hptserng@ntu.edu.tw(H.P. Tserng),Samuel@mail.ruentex.com.tw
(S.Y.L. Yin),rjdzeng@mail.nctu.edu.tw(R.J. Dzeng),wubin@cpami.gov.tw(B. Wou),
d93521006@ntu.edu.tw(M.D. Tsai),R94521711@ntu.edu.tw(W.Y. Chen). 0926-5805/$– see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.autcon.2009.05.005
Contents lists available atScienceDirect
Automation in Construction
but was also burdened with risks incurred from the owner or design
company[19]. RM, consequently, played a significant role for the
contractor. The risk manager is required to possess knowledge in order
to conduct risk management [18,26,33,16]. An important concern
regarding RM performance is how to reuse the knowledge base effectively. Accordingly, the ORM framework aims to enhance the RM performance through knowledge extraction and reuse.
In this study, the proposed ORM approach was found helpful for the selected contractor when conducting the project risk manage-ment. For the project manager (PM), this approach could be of
assistance in risk identification, analysis and response. In addition to
increasing RM effectiveness, the study verified that the project risk
ontology could be developed through acquiring tacit knowledge and extracting explicit knowledge from the organization. To summarize, in this study the ORM approach could support the contractor by
increasing effectiveness of RM workflow on the basis of implementing
knowledge management. 2. Literature review
2.1. Knowledge management research
Peter F. Drucker [31], the late master of management science,
attempted to clarify that knowledge had become the key asset to an organization in modern society. Later, the report from the
Organiza-tion for Economic Co-operaOrganiza-tion and Development (OECD)[28] —
knowledge, as embodied in human beings (as‘human capital’) and in
technology, had always been central to economic development —
more specifically indicated knowledge was essential to
Knowledge-based Economy. Therefore, knowledge asset had become a crucial
factor for organizational competitiveness[5,15], and was regarded as
highly important[29].
The crucial component of knowledge management (KM) is
mana-ging knowledgeflow that the organization needed[39,17]. Through the
application of the KM technique and IT tool, the procedures of
knowledge processing could be strengthened to help knowledgeflow
support the workflow operation in the organization[15,25]. Conversely,
Process-Oriented Knowledge Management (POKM) was conducted
using the analysis of workflow activities and knowledge requirements,
to integrate the appropriate knowledge management with the work-flow. Thus, the POKM approach could facilitate the knowledge flow in
each organization workflow and improve working performance because
of knowledge reuse[40,24,48].
For KM, knowledge processing procedures could be categorized into five steps: knowledge capturing, knowledge editing & validating, knowledge storing, knowledge sharing, and knowledge creating
[53,42]. However, knowledge management not only refers to knowl-edge-based management only, but it also covered the management of knowledge-creating processes, the management of person-to-person
knowledge exchange within the organization[49,17]. With reference to
Carrillo and Chinowsky's classification[29], the two KM types could be
divided into 1) the information technology centric (IT), and 2) the human resource centric (HRM). The HRM and IT types of KM can be regarded as two management models that correlate with tacit and explicit knowledge. For example, explicit knowledge correlates with the knowledge extraction of the enormous and complex project database, i.e., IT type. This presented a research issue about how to reuse the knowledge effectively. To address this concern, previous researchers
proposed several verified model to extract the construction
knowl-edge[10],[11],[51]. Thus, these studies foreground the importance of
knowledge extraction to the effective reuse of the knowledge. The available reused knowledge usually encompasses explicit and
tacit knowledge [15]. The tacit knowledge captured becomes a
problem of knowledge reuse[20,47]. To solve such a problem, many
studies suggested that explicit and tacit knowledge management should be conducted by combining the POKM approach and IT system
development during the workflow[5,12,44]. This solution adopted the
systematic process support to retrieve information and knowledge, and to record the users' decisions and tacit knowledge effectively during the working process. In summary, these studies have proposed
the verified models to support the capturing of the tacit knowledge by
embedding the KM mechanism into the workflow.
2.2. Information retrieval and ontology application in construction industry
Information retrieval can be defined as a technology relating to the
representation, storage, and organization of, and access to information
items.[8]. To simplify it, the goal of IR in the technical aspect is to
assist the users in quickly locating relevant and critical knowledge with much more ease. As the construction industry is characterized by its enormous, complex project data, how effective the knowledge dissemination and information sharing functions within the
organiza-tion are can provide high level value for the organizaorganiza-tion[4]. Thisfield
of technology is, therefore, recognized as an important aspect in the construction industry applications.
Ontology is a formal, explicit specification of a shared
conceptu-alization[46]. Specifically, ontology could be referred to the explicit
formal specification of the concepts in a specific domain and the
relations among them. Therefore, the two main elements of ontology are concepts and relations. Ontology could be relevant in the research field of artificial intelligence, representation of knowledge, semantic
web, system integration and problem solving techniques[1]. In the
construction management academia, ontology had been applied to knowledge representation, decision making, and information integra-tion[1,52,23,43].
The development of ontology requires a great deal of time, cost,
and expertise[23]. Previous efforts have focused only on a specific
topic[36,37,44,1]. Conversely, the most notable industry-wide efforts
that have been validated include the Industry Foundation Class (IFC)
and the e-Cognos project[36,37,52]. Most of the validated ontology
application models, however, have focused on a specific domain.
In previous studies, Kosovac et al.[6]assisted engineers with IR by
constructing a thesaurus. Lin and Soibelman[23]utilized a specific
ontology along with the algorithms for query expansion to enable users to retrieve and rank the information effectively. Moreover, semantic algorithms of the IR model were applied in the development
of several verifiable knowledge management systems (KMS) to
support architecture design and the organization of project
docu-ments [30,9]. In 2006, Rezgui [52] presented an application that
combined the semantic algorithms and the construction industry ontology to construct the automatic knowledge-feeding system based on the result of knowledge extraction. In summary, the previous
studies proposed several successful models which verified that IR
algorithms could support, retrieve and reuse of knowledge. Further-more, the IR algorithms could be applied with ontology.
2.3. Construction risk management
Tah and Carr's studies[18]indicated that RM procedure was widely
accepted as the chief role to affect RM. A good procedure design enabled a systematic and consistent approach to implement RM; hence, many studies were dedicated to research the RM procedure
[7,18,27]. Moreover, since risk management was important to project
performance, it had also been built into “A guide to the Project
Management Body of Knowledge (PMBOK guide)” framework
proposed by[32]. Through case studies, it was also proved that the
approach and tool mentioned in the PMBOK guide were influential to
project performance[22].
Another important RM research objective was to develop risk management approaches through experience and knowledge applica-tion. Hence, many studies discussed conducting risk management on
Analytical Hierarchy Process (AHP) and statistical techniques were adopted to transform professional experience and case knowledge into the RM model for achieving knowledge reuse. Based on these quantitative models, these studies found methods to enhance RM effectiveness via knowledge reuse. Furthermore, additional studies
were aimed at the specific project objective as the guideline to propose
the management model according to RM knowledge[50,11,7,21]. All of
these studies confirmed that RM knowledge was essential to RM
enhancement and RM is influential to project performance.
The previous RM investigation on construction industry pointed out that the most common RM problem during the project
construc-tion phase was the insufficient risk identification which tended to
cause the inadequate RM activity during the execution phase[22].
Another investigation concerning the British construction industry indicated that most companies would conduct risk management based on their previous experience, not on formal risk analysis
techniques due to time and knowledge insufficiency[2]. Subsequent
studies also concluded the same results [26]. These investigation
results, also, demonstrated that the typical RM problem consisted of complicated risk analysis techniques resulting in time and training
insufficiency leading to failed application of those techniques.
More recently, after construction companies began to adopt RM techniques to analyze risks, the following studies recommended that organizations should use qualitative or quantitative techniques to
analyze risk[41]. By analogy, qualitative, rather than quantitative,
techniques were widely accepted by the construction industries initially. In later years, when the construction companies became familiar with RM techniques, increasingly quantitative techniques
were adopted. Further, Wyk et al.[33]indicated that organizations
could identify the most appropriate RM application techniques by combination of qualitative, semi-quantitative, and quantitative tech-niques. In their study, another RM problem for the construction industry was proposed, as well: the organizations usually employed
more risk identification and evaluation than risk response and record.
By employing less risk response and record, verification for knowledge
reuse would be lacking and difficult to apply to the subsequent
projects. Thus, it could be concluded that construction RM shall be conducted based on the organization's requirements by using the formal and systematic RM technique. Furthermore, the RM knowledge database shall be renewed via audits and records to achieve the purpose of knowledge reuse and the enhancement of RM performance. 3. Research scope and methods
This study aimed to develop an ontology-based risk management (ORM) framework for the contractor. This study began with a review of the ontology development research. Based on literature review, the study adopted the following process to develop the project risk ontology
and ORM framework, including six steps: definition of scope, review of
domain authorities, extraction of important concepts, organization of
concepts into hierarchy, definition of the property of concepts, and
validation[1,43]. For the validation of the ontology, the competency
questions, expert survey and case study are the major validation
methods[1,43,44]. Moreover, the study adopted the IR algorithm to
develop the dynamic ontology extraction tool to supplement and update the ontology. Important in this study is while developing ontology, explicit and tacit knowledge for the effective reuse of knowledge must be included. Therefore, the study also included a literature review of the methods of explicit knowledge extracting and the tacit knowledge acquiring mechanism form the literature, and hence established the ontology development procedures.
By implementing the POKM model in RM workflow in the former
research[54], the selected contractor built the risk knowledge base. To
verify the project risk ontology and the ORM framework effectively, the study would concentrate on this risk knowledge base and the same contractor. Those involved in the case study were the risk
experts and senior engineers to provide a better understanding of the
risk profile and risk knowledge. In this case study, expert survey and
the case study were adopted to verify the project risk ontology. In order to verify this risk ontology, the study conducted expert survey and several expert workshops to discuss and modify the ontology. Subsequently, the feasibility of the ORM framework and the project
risk ontology would be verified via 5 actual project demonstrations.
Finally, the ORM framework and case study results would also be validated via extensive discussion with domain experts.
4. Development of ORM framework
The development of the ORM framework included four primary domains: (1) the development of project risk ontology; (2) the extraction of the risk knowledge base; (3) the establishment of the ORM approach; and (4) the dynamic ontology extraction tool. Based
on the verified model of the ontology development and application in
the construction industry, the ontology development process includes six main steps. The research process included six steps as illustrated in
Fig. 1. The conceptual model of the proposed framework was shown in
Fig. 2. In this model, the project risk ontology played the most vital role of the entire framework. The ORM approach was developed based on the characteristic of project risk ontology and the application
problems of RM workflow. The knowledge extraction model
encom-passed both explicit and tacit knowledge. As previously indicated, the developing of ontology involves much expertise and many case studies. Moreover, the involvement of domain expertise in intensive interviews and the iterative development of procedures are crucial for
ontology development[55,38].
4.1. Definition of scope
The ORM framework was established to enhance RM performance
through ontology development and application. The RM workflow
and demand knowledge analysis are essential to the ontology development process. As the aforementioned studies indicated,
knowledge record was influential to the subsequent project during
RM workflow, since complete knowledge flow is necessary for the
knowledge base.
For the ontology development, activity analysis is elemental to identify the scope. This study performed the activity analysis based on the literature review. As the previous studies showed, that the industry
widely accepted RM as a formal and systematicflow composed of five
steps: risk plan, risk identification, risk analysis, risk response, and risk
monitor and feedback[18,10,41,33]. Accordingly, this RM workflow
was the basis for the developing the ontology to propose a more generalized solution to the construction practitioners.
Analysis of knowledge demand of RM workflow shapes and
defines the scope. This study also performed activity and knowledge
interaction analysis of RM workflow. According to Tah and Carr's study
[18], the standard methodology such as IDEF0 could be used to model
the process and informationflow of RM workflow. Therefore, this
study adopted IDEF0 to model the activity and knowledge interaction
among the RM workflow as illustrated inFig. 3. InFig. 3, each activity
box shows the main activity of the RM framework. Moreover, this
diagram also showed the knowledgeflow and its interaction with RM
workflow. InFig. 3, four arrow classes are illustrated in each activity
box: Input arrow, Output Arrow, Control Arrow, and Mechanism
Arrow. This study analyzed the demand knowledge flow of each
activity and its arrows classes, as illustrated inTable 1. As presented in
Table 1, there are 2 main RM knowledge types: (1) explicit knowledge: RM knowledge, including historical risk knowledge, RM template and
predefined RM approaches; and (2) tacit knowledge: RM knowledge
and experience that resided with individuals. In summary, the analysis
of process domain and knowledge domain not only defined the scope,
but also identified the knowledge resource of ontology development.
4.2. Requirement collection
Previous studies alluded to the difficulties in applying formal RM
procedures. The time consumed and the complexity involved could lead to failed application of RM techniques. Moreover, the application of existing qualitative and quantitative techniques suffered for the
insufficient knowledge and effectiveness. Therefore, the ORM
frame-work was proposed to decrease the complexity of RM frame-workflow and
increase effectiveness.
To develop the ontology requires combining both explicit and tacit
knowledge for reuse. The knowledge in RM workflow needed further
analysis. Based onTable 1, this study designed different knowledge
gathering means for different RM knowledge types, as illustrated in
Table 2. As indicated inTable 2, the main RM knowledge type that can be acquired within the organization is explicit knowledge. Contrarily, the
tacit knowledge (RM experience) which is demanded throughout RM
workflow is difficult to capture and reuse. Moreover, the lack of tacit
knowledge often presents problems in RM. Therefore, the means designed to capture RM experiences among different scenarios must be of equal vigour in the ontology development. Moreover, for
implement-ing the ORM framework,Table 1would be applied as the risk knowledge
check list andTable 2serves as risk knowledge capture means.
4.3. Review of domain authorities
The review processes for domain authorities include a review of
existing classification systems and ontology. In the risk classification
domain, previous researches proposed inclusion of several classi
fica-tion systems. However, most of those systems concentrated on specific
RM domains, i.e., BOT project, scheduling, and safety management
[7,45,10]. In contrast, the industry wide risk classification domain, the
hierarchical risk-breakdown structure (HRBS) developed by [18],
would be relevant to this study. Thus, this study adopted the HRBS
classification as the underlying ontology. As this study selected a
specific domain of ontology development and application, the industry
wide ontology such as IFC and e-Cognos were only adopted as the referential source for the project risk ontology development.
The expert workshop is useful for this step to provide the basis of ontology development. In order to propose a more feasible ontology
for the following verification, domain experts from the selected
contractor were invited to modify the underlying classification. There
were 3 expert workshops held with 25 participants to discuss the
underlying classification and to evaluate other classifications from the
literature review in 2006. And the results provided the basic taxonomies for the following ontology development.
4.4. Extraction of important concepts
Based on the basic taxonomies, the extraction of important concepts plays the role of collecting the concepts for the ontology development. Previous studies adopted a variety of methods to extract important concepts from the target domain including a review of an existing taxonomy, a review of the literature, an analysis of a sample document,
and a conceptualization of IDEF0[38,43,1]. For implementing the ORM
framework, these concept extraction means could also be flexibly
adopted for the contractors.
As discussed earlier, the development of ontology required a lot of the expertise. Therefore, in this step, the project risk expert and senior manager interview or workshop were necessary for the major
modification of the preliminary set of risk concepts.
4.5. Organization of concepts into hierarchy
Following the completion of the project risk concepts extraction
process, the classification framework could be classified into risk
knowledge types. In this step, the major class and class hierarchy of
the ontological framework could be identified based on the literature
review and case studies[1]. Previous study also suggested that both
classification and the balance between depth and coverage[38]could
change the development result. However, there is no specific standard
for the validation of ontology[36]. Thus, the involvement of domain
expertise in intensive interviews and the iterative development of procedures were the most important parts in this step.
In this step, the risk ontology framework composed of risk class,
subclass and indexes as shown inFig. 2was developed. A risk subclass
could also view as risk events. The remaining risk indexes were referred to certain characteristics/instance of risk events.
4.6. Establishment of the ORM approach
As discussed earlier, regarding the collected requirement, the ORM
framework aimed to decrease the complexity of RM workflow,
increase effectiveness, and develop ontology by combining both explicit and tacit knowledge. The result of activity and knowledge
interaction analysis would advance an adequate approach to fulfil
these requirements. The qualitative and quantitative risk analysis of
RM workflow can provide strong information about project risk.
Moreover, previous studies also mentioned that such procedures are
Table 2
RM knowledge and KM means analysis of RM workflow.
RM knowledge Knowledge type Description Knowledge capture means Predefined RM approaches Explicit Predefined RM approaches could be: risk categories,
RM templates, risk assess methods, roles and responsibilities.
Documented knowledge acquisition/compilation mechanism RM experience Tacit RM experience could be risk management experience and
related experience to certain risk throughout project life cycle.
Interview, project meeting(knowledge community), questionnaire Risk categories Explicit Risk categories are predefined based on the risks historical data
though classification.
Construction of knowledge classification/ map
Historical risk knowledge Explicit Related project document, risk report, and RM plan. Documented knowledge acquisition/compilation mechanism Risk checklist Explicit Risk checklist is predefined based on the risks historical data. Documented knowledge acquisition/compilation mechanism Table 1
Activity and knowledgeflow analysis of RM workflow.
RM activity Arrow type Description Demand RM knowledge Risk management planning Input arrow Review/confirm of project related • Predefined RM approaches
• Enterprise environment • RM experience • Organizational process assets • Risk categories • Project goals and objectives • RM plan template Mechanism arrow Based on initial RM knowledge, project manager (PM) and
RM team conducts RM planning meeting and analysis.
• RM planning experience Risk identification Input arrow Based on the RM plan and strategies, PM reviews and
gathers the organizational process assets.
• Predefined RM approaches • RM experience
• Historical risk knowledge Mechanism arrow PM uses the appropriate risk identification techniques to
identify risks might affect project.
• RM experience • Risk checklist • Risk categories • Historical risk knowledge Qualitative risk analysis Input arrow Based on the identified risks and risk profile, PM reviews
and gathers related risk knowledge.
• RM experience • Historical risk knowledge Mechanism arrow Based on predefined approaches, PM conducts risk assessment. • RM experience
• Historical risk knowledge Quantitative risk analysis Input arrow The probability and impact scale of identified risks.
Mechanism arrow To analyze each risk and its consequence on project objectives. • RM experience • Historical risk knowledge Risk response planning Input arrow • Priority of risks and probability of achieving the project objectives
• RM plan and strategies.
Mechanism arrow PM plans and adopts Strategies and responses for uncertainties to reduce threats of risks
• RM experience • Historical risk knowledge Risk monitor and control Input arrow • Risks profile and its response • RM experience
• RM plan and strategies. • Historical risk knowledge • Residual and secondary risks
Mechanism arrow PM tracks the effectiveness of responses. Also, PM identifies and assesses new risk. Last, status meetings are held to review the project risks.
• RM experience • Historical risk knowledge • Risk checklist
difficult for contractor to apply. Thus, this study would propose the solution to integrate both these procedures.
During the qualitative risk analysis process, the risk matrix calculation helped the risk manager to analyze the risk impact scale as to the schedule, cost and quality, and to distinguish priority via probability and impact data elements that lead to rating the risks as
low, moderate, or high priority[32]. Furthermore, the most important
component of the quantitative risk analysis process is the risk priority. To summarize, the goal of both of these procedures is to avoid the project manager from assessing individual risk events only, based on previous experience and intuition by giving numeric values. Therefore, by utilizing the project risk ontology, the study proposed another approach to conduct the numeric analysis of project risk.
As mentioned earlier, the risk classification could aid the project
manager to identify the possible risks. Therefore, the project risk ontology could support to identify the project risks distinguished by the risk categories, subclasses and indexes. Consequently, the
amounts of identified risk could represent the quantities of project
risk degree. However, the impact scales among these risks can fluctuate. Thus, the weights of each risk class and subclass were required to quantitatively measure the severity of project risk degree.
Thefirst step in establishing the risk numeric analysis model was to
set up risk class and subclass weight. Based on the risk classification, the
following step was to set the weight of each risk subclass and risk class to
the influence of each project goal, i.e., cost, time, and quality. To capture
the experts' experience and assessment on risk categories and subclasses, this study found the Analytical Hierarchy Process (AHP) was ideal to this step. The AHP method assists in making both qualitative and quantitative decisions. From the literature, the AHP method has been adopted widely in decision making among construction industry
[45,11,1]. Therefore, this study adopted the AHP questionnaire to set the weights of each risk class and subclass by acquiring the tacit risk knowledge.
To represent project risk by numerical rating, a risk quantitative assessment method was required. Accordingly, project risk evaluation criterion was established, according to the risk indexes for each risk
event established in theSection 4.5. Furthermore, these evaluation
criterions could also help PM to identify the possible risk when it was used as a risk checklist. Thus, this solution could integrate the
proce-dures of risk identification and risk analysis.
Following the development of risk indexes, the ORM framework was
built as illustrated inFig. 2. By using the evaluation framework above, the
user could evaluate severity of each risk event with reference to
evaluation criterions (seeFig. 2). Therefore, each risk event would have
numerous evaluation criterions, and the total risk score of each risk event would be the total number of the criterions that are met. The user
could review and obtain the risk score of each individual risk event
through the evaluation criterions, as shown in step 1 in Fig. 2.
Subsequently, the user could calculate severity of each risk event according to weight obtained from AHP, as shown in step 2 in the Diagram. The calculation formula for the risk severity was as follows:
Risk value of risk event =X
n
i = 1
Risk scores of
risk event⁎ weight of risk event
ð1Þ
Taking External risk categories as the example, the calculation
formula for ERVt is ERVt=P
4 i = 1
ERVi =P
4 i = 1
ERSi4ERWi (ERS was risk
score of individual risk event; ERW was risk weight of individual risk event; ERVt was the overall risk score for all risk categories).
More importantly, the user could adopt the weight of each risk class to calculate the overall risk score for the entire project, that is, Project Risk Index (PRI), as shown in step 1, and to evaluate the overall risk for the project, accordingly. The calculation formula for PRI was as follows:
PRI =X
8
i = 1
Risk value of risk c; ass⁎ weight of risk class ð2Þ
Where:
Risk value of risk class =X
n
i = 1
Risk scores of
risk event⁎weight of risk event
For the aspect of PRI application, this risk numeric analysis model could support PM to assess the overall risk of the project and risk distribution via PRI value. Moreover, the contractor could also obtain project risk situation through the analysis of time series on overall PRI that was calculated from the project's inception. The contractor could accumulate and calculate PRI values of different types of projects to construct a standard for aiding in decision-making on RM in the future. Therefore, the organization could adopt the calculation of PRI
value and the comparison of the historicalfigures to monitor and
evaluate the overall risk. To summarize, the project manager could gain an understanding of the overall risk value of project risk and risk situation of individual risk type via the model. Therefore, the model
Fig. 4. The keyword view of vector model.
could not only help the project manger conduct risk quantitative assessment, but calculate risk and decide the risk response priority, as well. Moreover, by combining explicit knowledge extracted and the expert's tacit knowledge, the ORM framework could improve the risk knowledge reuse.
4.7. Establishment of the dynamic ontology extraction tool
From the literatures, the development of ontology requires a great deal of time, cost, and expertise. In this study, the project risk ontology development also involved many expert interviews and document analysis. Therefore, the update of ontology became the critical research issue to maintain the usability of ontology. To address this issue, the study developed the dynamic ontology extrac-tion tool to supplement and renew the ontology. Thus, the prototype
of the IR algorithm-based dynamic risk profile analysis tool was
developed.
IR models were developed to provide the analytical ability to
measure document relevancies[23]. Through using IR algorithms, it
could help the document summarization by extracting important
terms[52]. Hence, the document could represent as a set of important
term vectors. In IR research domain, these important terms are also called a set of semantically relevant keywords. Moreover, the document relevancies were calculated based on these important terms. Subsequently, the relevancy of documents could be decided by the Vector model. In the Vector model, two documents would transfer into two individual vectors in virtual space.
In this study, the proposed dynamic ontology extraction tool was focused in the extraction of important risk concepts. Therefore, the study changes the document relevancy calculation model from the document
view into the keyword view, as illustrated inFig. 4. In the keyword view
vector-space, this study adopted a document vectors set to identify keywords. The keyword relevancies can be decided by the same Vector model. Hence, the relation between risks can be determined by this
Vector model. Through the timely analysis of IR algorithm, the ontology extraction tool can support the dynamic update of ontology.
In this Vector model, the similarity between two keywords (risk concepts) can be measured by the closeness of two keyword vectors. From the literature, the similarity can be calculated by cosine value of
two keyword vectors: kwiand kwj, as in Eq.(6) [14]. Eqs. (3)–(5)
define the term weighting calculation of each keyword. Moreover, the
coefficient 0.5 was used to normalize the term frequency proposed by
Salton and Buckley[13].
Wdj; kw= 0:5 + 0:5 × tfdj; kw × idfkw = 0:5 + 0:5 × frekw max fre × log N nkw ð3Þ tfdj; kw= frekw max fre ð4Þ idfkw= log N nkw ð5Þ similarity →kwj;→kwk = cosine →kwj;→kwk = Pn i = 1 Wdi; kwj× Wdi; kwk ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn i = 1 Wdi2; kwj s × ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn i = 1 Wdi2; kwk s ð6Þ Where:
Wdi, kwis the term weight of keyword kw in document di; frekwis
the frequency of each keyword; max fre is the maximum frequency of
keyword kw in document di; P
sj i = 1
fre is the total frequency of all keywords in each document; N is the total document numbers of
knowledge base; nkw is the total numbers of document in which
certain keyword appears.
Based on this IR algorithm, this study developed the prototype of ontology extraction tool. Through the deployment of IR algorithms, this tool can process text as follows: (1) document cleaning; (2) keyword extraction; and (3) selection of the important keyword (risk concept). In this study, the document cleaning work included
the removing of stop-words and affixes. Moreover, this tool replaced
the synonyms with the same term in the analyzed text based on the
thesaurus developed in theSection 4.4. Through these steps, the tool
can extract the most important risk concept based on the correct frequency of term appearances. And the prototype (Prototype A) was built up by integrating the Chinese extraction module, the document cleaning module, the similarity calculation module, and a user interface.
The ontology extraction tool was designed to dynamically extract important risk concepts and hence supplement the project risk
ontology. Therefore, the terms extraction validity must be confirmed.
In this step, the study adopted the valid lexicons extracted rate to test the validity of the ontology extraction tool. Moreover, the study used the web record to test the terms extraction validity. The equation for the terms extraction validity was as below:
Terms Extraction Validity¼Valid Lexicons ExtractedTotal Lexicons Extracted ð7Þ
In the validation for prototype, the study randomly selected 100 web records to measure the performance of the tool. Moreover, the study used 2 term similarity threshold, including: low similarity threshold (0.70) and high similarity threshold (0.90) to test the terms
extraction validity. The results of this testing was illustrated inFig. 5.
From the testing, the valid lexicons extracted rate was mostly over 80% for both similarity thresholds. Moreover, the high similarity threshold could derive better extraction performance. Through the testing of the prototype, the performance of the ontology extraction tool was proved. 5. Case demonstration
To validate the effectiveness of the ORM framework, this study implemented the ORM framework for actual construction projects. The case study of this study was derived from the Ruentex Construction & Engineering Company established in 1977 in Taiwan, which was a rare enterprise with a strong integration-ability of horizontal and vertical levels. There are 500 employees in the entire enterprise. The business scope includes construction, design, precast, mechanical, interior design, building security service, elder caring and nursing home, real-estate, etc. This general builder began to develop its management information system in 1983, and began an evaluation on the ERP implementation in June 2001. Moreover, the selected contractor had implemented the PMBOK project management approach and the web-based RM system
as the ICT solution of the project RM workflow in 2004. The scope of the
implementation included: (1) the development of project risk ontology; (2) the extraction of the risk knowledge base; (3) the implementation of
the proposed ORM approach; and (4) the modification of the dynamic
ontology extraction tool. Subsequently, this study conducted an
inter-view with the selected contractor's project managers to confirm the
Table 3
AHP weights of risk categories and risk subclasses for the selected contractor.
Risk category Risk weight Risk weight of each risk subclass
Site condition (SC) 0.078 Subclass 1: 0.173; Subclass 2: 0.115; Subclass 3: 0.345; Subclass 4: 0.368
Owner–contractor agreement (OA) 0.103 Subclass 1: 0.102; Subclass 2: 0.070; Subclass 3: 0.170; Subclass 4: 0.215; Subclass 5: 0.231; Subclass 6: 0.212 Owner condition (OC) 0.185 Subclass 1: 0.220; Subclass 2: 0.233; Subclass 3: 0.118; Subclass 4: 0.204; Subclass 5: 0.226
Subcontractor condition (SuC) 0.104 Subclass 1: 0.280; Subclass 2: 0.260; Subclass 3: 0.135; Subclass 4: 0.325
Project execution (PE) 0.189 Subclass 1: 0.122; Subclass 2: 0.226; Subclass 3: 0.311; Subclass 4: 0.228; Subclass 5: 0.112 Project preparation and planning (PP) 0.147 Subclass 1: 0.162; Subclass 2: 0.224; Subclass 3: 0.333; Subclass 4: 0.168; Subclass 5: 0.112
Contracting and administration procedure (CA) 0.121 Subclass 1: 0.129; Subclass 2: 0.128; Subclass 3: 0.196; Subclass 4: 0.293; Subclass 5: 0.163; Subclass 6: 0.090 External risk (ER) 0.073 Subclass 1: 0.128; Subclass 2: 0.108; Subclass 3: 0.439; Subclass 4: 0.324
Table 4
Consistency ratio (C.R.) of each risk category of the selected contractor.
Risk category C.R.
Site condition (SC) 0.01
Owner–contractor agreement (OA) 0.01 Owner condition (OC) 0.012 Subcontractor condition (SuC) 0.004 Project execution (PE) 0.013 Project preparation and planning (PP) 0.022 Contracting and administration procedure (CA) 0.037
feasibility and modification of the ORM framework application. More-over, this study evaluated the ORM framework in terms of the management-level interview results. Finally, the study also held an expert workshop to discuss and validate this framework.
5.1. Definition of scope
Through the implementation of the POKM model and the PMBOK project management approach, the contractor knowledge base in the
previous research provides a strong foundation for the knowledge extraction and ontology development. Therefore, this study adopted this knowledge base as the resource for the knowledge extraction.
As discussed earlier, analysis of knowledge demand of RM workflow
shapes and defines the scope for the ontology development. Therefore,
the study conducted the interview with the RM team of the selected contractor to identify the possible knowledge source of ontology
development based in the Table 2. From the interview, the study
identified the risk data base of the web-based RM system and the risk
classification would be the main sources for the project risk ontology
development.
5.2. Requirement collection
For the selected contractor, although it had implemented the KM in
the RM workflow, the entire RM workflow still lacked an assessment
model for risk quantitative analysis to support the RM decision making. In addition, the selected contractor lacked an integrated risk monitor and assessment tool. For these objectives, this study assisted the selected contractor in implementing the ORM framework to overcome the aforementioned problems.
5.3. Extraction of important concepts for risk ontology
As discussed earlier, a variety of methods could be used to extract important concepts from the target domain. In this step, this study adopted document analysis as the major concept extraction method.
To establish the knowledge database for concept extraction, this study gathered risk knowledge on construction projects. First, this study conducted a survey of typical risk events in each functional departments of the selected contractor, i.e., procurement, construc-tion planning. In this step, 132 risk types were built from 8 funcconstruc-tional
departments. Second, this study would assemble the risk profiles of
the risk knowledge base. The risk profiles consisted of risk events,
including: risk event description, cause, consequence, response, risk status, and qualitative analysis. Subsequently, all project managers (PM) and functional department managers within the organization
Fig. 8. Screen shot of the dynamic ontology extraction tool.
Table 5
The similarity calculation result of the test retrieval.
Related risk concept Similaritya
Affect the schedule 0.8350876
Architect 0.844618
Corrective action 0.9446967
Customizing 0.8350876
Drawing manpower allocation 0.9632335 Drawing progress 0.978044 Limited experience 0.7982844 Long review process 0.9446967 New construction method 0.8280177
Owner review 0.7710148
Require detailed construction drawing for new technique 0.8493434 The rubber shock isolator installation 0.8160871 Variation orders 0.7389783
a
were requested to conduct the survey of risk events concerning their current respective projects. Through this procedure, the initial
knowledge database of 115 risk profiles among 13 projects was
established while the risk experiences were compiled and clearly documented. After converging the relevant risk knowledge and
information, the analysis of risk profiles was performed. Hence, a
preliminary set of concepts was established. The set of approximately 500 concepts emerged.
Following the completion of the preliminary set, the study refined
it through expert interviews. In this study, 5 project managers with an average of 21 years of working experience from the selected contractor were interviewed. These managers were chosen because they are the RM experts in the major project types in this selected contractor. In this case study, the major project types are high-tech
factory, residential building and office. Following the interviews, the
preliminary set was refined again into approximately 200 concepts. In
addition, a thesaurus of project risks was also established for a further information retrieval operation. Thus, in this stage, the extraction of explicit knowledge was completed.
5.4. Development of the project risk ontology
Following the basic taxonomies of ORM framework, the risk
concepts could be organized into risk classification. By RM operation
scenario, the risk classification framework could be applied to risk
identification; therefore, the basic risk taxonomies needed the
modification for the selected contractor. To address this issue, the
study collaborated with the RM team to modify the basic risk taxonomies. Moreover, in this step, the same respondents (those
five managers mentioned inSection 5.4) were interviewed again. To
further analyze project risk, this study established a cause-and-effect diagram for the purpose of project risk analysis. In this phase, as
demonstrated inFig. 6, this study determined 8 risk categories, 39 risk
subclasses and 195 risk indexes in the third tier.
Furthermore, to the issue of the iterative development of ontology, the study also held a senior managers' workshop, consisting of the heads of each department and senior managers to discuss the
hierarchical risk classification. In this workshop, 6 senior managers
or engineers with an average of 30-years of working experience
reviewed and modified this hierarchical risk classification.
5.5. Implementation of the ORM approach
Thefirst step in establishing the risk numeric analysis model was
to set up risk class and subclass weight. Therefore, this study designed the AHP questionnaire to conduct full pair-wise comparison on all risk categories and all risk subclasses in each risk category. The hierarchy
framework of the questionnaire was shown inFig. 6.
The questionnaire was designed for those project managers who had been working for the selected contractor for over 15 years and had profound experience with the construction industry. A total of 25 questionnaires were disseminated to the subjects: 20 completed questionnaires were returned from the subjects, and used in this study. Because the questionnaire was complicated, the subjects' responses to the questionnaires were answered by telephone inter-view or face-to-face discussion. After the completion of question-naire analysis, this study set AHP weights for each risk category and
risk subclasses of each risk category, as illustrated inTable 3.
Next, Reliability Analysis was conducted to ensure the reliability of the questionnaire results. For reliability analysis, Consistency Analysis was used to judge the reliability of each expert's assessment based on
Consistence Ratio (C.R.) when C.R.b0.1.This study used Consistency
Analysis to analyze each hierarchy and overall hierarchy. The result represented Consistency Index of overall hierarchy (C.I.H.) was 0.0278 and consistency ratio of the overall hierarch (C.R.H.) was 0.0255. It
could be found that C.R.H.b0.1, therefore, consistency of overall
hierarchy was acceptable. Otherwise, consistency checking for each
hierarchy stated the same result, as presented inTable 4.
Based in the survey results and the project risk ontology, the project risk quantitative analysis model of the ORM framework was
built, as illustrated inFig. 7. Subsequently, the project managers could
calculate the risk values for risk events, classes and overall risk score
Table 6
Test retrieval results for Prototype B. No. Query Analyzed
relevant risk profile Relevant extracted risk concepts Extracted risk concepts Relevant risk concepts Precision (%) Recall (%) 1 Construction diagram 6 17 19 21 89 81 2 Construction Site Condition 13 15 18 18 83 83 3 Owner Condition 16 29 35 34 83 85
for the entire project (PRI). In particular, this study, along with the selected contractor, conducted tacit knowledge acquisition within the organization via the result of this AHP questionnaire. The outcome could not only establish the base for quantitative assessment model, but also help the project manager judge the management priority of risk events in the project.
5.6. Implementation and evaluation of the dynamic ontology extraction tool Among the ORM framework, the ontology extraction tool was designed to dynamically extract important risk concepts and hence supplement the project risk ontology. To address this issue, the study collaborated with the RM team to design the using scenario of the tool. According to the characteristics of project risks, the risk classes and subclasses are static, while the risk indexes are much more dynamic and require the continuous update by the RM team. Thus, for the selected contractor, the tool was designed to retrieve these concepts based on the risk subclass and any important risk concept. The prototype (Prototype B) for the selected contractor was built up by the
Prototype A (seeFig. 8).
As shown in Fig. 8, the study used the query “construction
drawing” to retrieve the related risk concept in this test retrieval.
Subsequently, the related risk concepts were retrieved by this tool. Moreover, the relation between retrieved risk concept and risk
query could be identified through the similarity calculation as
illustrated inTable 5. In this test retrieval, the similarity threshold
was set to 0.70 (low similarity threshold) to avoid the missing of important risk concepts. And such threshold number could be
modified through further test trial by the selected contractor. In this
test retrieval, the risk concept“Drawing manpower allocation” was
highly related to the query. And the detailed calculation was shown as below:
similarity
→
queryj;→
drawing−manpowerk! = Pn i = 1 Wdi; kwj× Wdi; kwk ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn i = 1 W2 di; kwj r × ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn i = 1 W2 di; kwk r =4:9665495:15612 = 0:9632335
According to this test retrieval, the risk concept“drawing
man-power allocation” was the most important risk in the timely project
risk nature of the case study company. Moreover, to validate the extraction performance of risk concept, 3 test retrievals were conducted to provide the measurement of the extraction validity.
Table 6listed the extraction performance of the test retrievals. In these
tests, 35 risk profiles were retrieved from the risk knowledge base. The
knowledge base composed of 115 risk profiles was built up from 2005
to 2006. Due to this tool focused on extracting risk concepts than the
retrieval of relevant risk profiles, the precision and recall rate[8]were
adopted and modified to evaluate the extraction performance of the
proposed tool as below:
Precision =Relevant Extracted Risk Concepts
Extracted Risk Concepts ð8Þ
Recall =Relevant Extracted Risk Concepts
Relevant Concepts ð9Þ
From the testing, the average precision and recall rate were over 80% and accepted by the selected contractor. Furthermore, the other
dynamically extracted risk concepts couldfilter by similarity
thresh-old or expert judgements. Afterwards, the retrieval result could provide the possible risk indexes to supplement the original risk
indexes. Thus, this dynamic information could help the identification
of possible risk and supplement of project risk ontology. To
summarize, the mechanism to use this tool is illustrated inFig. 9.
5.7. Filed trial evaluation
After the implementation of the ORM framework, the study
verified the ORM framework through 5 construction projects of the
selected contractor including 2 high-tech factories, 2 residential
buildings and 1 office building. These projects were the most typical
project of the selected contractor. The project managers of these 5
projects used the ORM framework ranging from risk identi
fica-tion, risk analysis, to conducting further project risk analysis and planning.
Table 7listed the testing results of thefiled trials. Following the adoption of this ORM framework, the project managers pointed out
that project risks were identified through this framework accurately.
Moreover, they also mentioned the risk score and severity could be determined simultaneously by applying the AHP weights of risk classes and subclasses. Hence, this framework could facilitate risk
identification, analysis, and response procedures. The field trial results
proved that risk values and distributions calculated from the ORM
framework could reflect the project status based on PRI calculation.
Through the implementation of the ORM approach, these project managers could employ the accurately responses with the correct priority setting. Therefore, the project risks were effectively handled. The Project RM performance measurements were demonstrated in
Table 8. Moreover, in addition to proving that the model could help PM
identify and monitor project risk, the outcome also verified the
effectiveness of the project risk ontology development model. Furthermore, the study held a domain expert workshop to discuss the ORM framework and the test case studies. In this study, 6 domain experts with an average of 25-years of working experience were interviewed. The workshop began with 20 min presentation about the ORM framework development and the case study, and the experts
Table 7
Test results of the project risk quantitative assessment model. Project Project type Contract budgeta
(US dollar) Risk value of each risk category PRI Evaluation
A High-tech factory $10,303,030 SCVt 1.909 OCVt1.524 OAVt 2.318 SuCVt 2.695 2.329 Accurate identification and analysis of project risk PEVt 2.995 CAVt 1.74 PPVt 2.77 ERVt 2.673
B High-tech factory $696,969 SCVt 3.197 OCVt0.118 OAVt 0.482 SuCVt 1.195 1.035 Accurate identification and analysis of project risk PEVt 0.228 CAVt1.363 PPVt 1.712 ERVt 1.782
C Office $16,515,151 SCVt 1.403 OCVt0.555 OAVt 0.312 SuCVt 0.27 0.861 Accurate identification and analysis of project risk PEVt 0.808 CAVt0.971 PPVt 1.326 ERVt 1.694
D Residential building $4,490,606 SCVt 2.347 OCVt1.123 OAVt 2.495 SuCVt 1.56 1.567 Accurate identification and analysis of project risk PEVt 1.119 AVt 1.677 PPVt 1.88 ERVt 0.911
E Residential building $42,606,060 SCVt2.8757 OCVt 0.57 OAVt 0.964 SuCVt 0 1.092 Accurate identification and analysis of project risk PEVt 1.449 CAVt0.715 PPVt 1.45 ERVt 1.235
a
were then asked to provide the suggestion and inquiries. The major suggestions are summarized below:
1. The ORM framework could reduce the complexity of common RM
workflow effectively, especially via the risk analysis stage of the
process. Moreover, inexperienced engineers could adopt the framework easily when conducting decision-making.
2. The ORM framework could help contractors to transfer organiza-tion knowledge to handle the project risks better. Moreover, the tacit knowledge could record and apply to the subsequent projects. 3. The ORM framework and the ontology development model are
solid and could apply in other RMfield such as disaster handling
and earthquake risk management. They also suggested future
research could work in thesefields to extend the application field of
the proposed framework. 6. Conclusion
This ORM framework aimed to verify the project risk ontology could
enhance RM workflow performance within the construction
organiza-tion by case study, and hence decrease risk impact on the project. Through the implementation of the ORM framework and the ontology development model, the selected contractor could integrate knowledge reuse in the RM operation. Consequently, based on knowledge accumulation and reuse, the aforementioned RM problems could also be overcome and hence support the RM decision making.
The study attempted to propose a knowledge extraction model that was applicable to construction project risk ontology development. Through literature review, the study analyzed the relevant approaches proposed by previous studies and the characteristics of construction
RM. After considering the requirements of construction RM workflow,
the study adopted the expert interview and AHP approach to establish the ontology. Moreover, the IR algorithm-based ontology extraction tool was developed to supplement the ontology. To summarize, this study proposed 1) project risk ontology development model, and 2) risk knowledge extraction model of ORM framework.
By case study, this study proved that the ORM framework could help the contractor conduct knowledge reuse during RM procedures to decrease risk threats to the project. Simultaneously, the outcome also
confirmed the effectiveness when project risk ontology was applied to
construction RM workflow. Through knowledge extraction and reuse,
the organization could not only manage project risk more efficiently,
but, also disseminate the organization risk knowledge effectively.
Moreover, the contractor could decrease RM workflow complexity and
increase organization RM effectiveness via this ORM framework. In addition, the study validated that through the mechanism of extracting explicit knowledge and acquiring tacit knowledge, these
two types of knowledge could be combined and applied to workflow.
Based on ORM framework, the organization could adopt the approach
to enhance workflow performance. Moreover, different from the
previous deficiencies, the approach could assist the organization in
acquiring and reusing tacit knowledge when applying KM.
With regard to application of the ORM framework and the ontology development model to other construction practitioners, the model was proposed with reference to typical RM standard procedure and knowledge application scenario. Moreover, the ORM
framework was validated through case study verification and
extensive domain expert interviews. Therefore, the construction practitioners could apply this ORM framework to establish project risk ontology and enhance RM performance through knowledge reuse. References
[1] A. Pandit, Y. Zhu, An ontology-based approach to support decision-making for the design of ETO (Engineer-To-Order) products, Automation in Construction 16 (6) (2007) 759–770.
[2] A.S. Akintoye, M.J. MacLeod, Risk analysis and management in construction, International journal of project management 15 (1) (1997) 31–38.
[3] A.S. Ang, D.D. Leon, Modeling and analysis of uncertainties for risk-informed decisions in infrastructures engineering, Structure and Infrastructure Engineering 1 (1) (2005) 19–31.
[4] A.R. Mohamed Eshaq, P. Karboulonis, Design considerations for the design of an advanced VR interface for knowledge management and its relevance to CAD, Automation in Construction 12 (5) (2003) 501–507.
[5] B.H.W. Hadikusumo, Steve Rowlinson, Capturing safety knowledge using design-for-safety-process tool, Journal of Construction Engineering and Management 130 (2) (2004) 281–289.
[6] B. Kosovac, T.M. Froese, D.J. Vanier, Integrating heterogeneous data representations in model-based AEC/FM systems, Proc. CIT '00, vol. 2, IABSE, 2000, pp. 556–567, Reykjavik, Iceland.
[7] B. Mulholland, J. Christian, Risk assessment in construction schedules, Journal of construction engineering and management 125 (1) (1999) 8–15.
[8] B.Y. Ricardo, R.N. Berthier, Modern information retrieval, Addison Wesley, New York, 1999.
[9] C.H. Caldas, L. Soibelman, J. Han, Methodology for the integration of project documents in model-based information systems, Journal of Computing in Civil Engineering 19 (1) (2005) 234–243.
[10] D.K.H. Chua, Y.M. Goh, Incident causation model for improving feedback of safety knowledge, Journal of Construction Engineering and Management 130 (4) (2004) 542–551.
[11] F.M. Arain, L.S. Pheng, Knowledge-based decision support system for management of variation orders for institutional building projects, Automation in Construction 15 (3) (2006) 272–291.
[12] G. Carter, S.D. Smith, Safety hazard identification on construction projects, Journal of Construction Engineering and Management 132 (2) (2006) 197–205. [13] G. Salton, C. Buckley, Term-weighting approaches in automatic retrieval,
Informa-tion Processing & Management 24 (5) (1988) 513–523.
[14] G. Salton, Automatic Text Processing: The Transformation, Analysis, and Retrieval of Information by Computer, Addison Wesley, New York, 1989.
[15] H.C. Tan, P.M. Carrillo, C.J. Anumba, N. Bouchlaghem, J.M. Kamara, C.E. Udeaja, Development of a methodology for live capture and reuse of project knowledge in construction, Journal of management in engineering 23 (1) (2007) 18–26. [16] I. Dikmen, M.T. Birgonul, C. Anac, J.H.M. Tah, G. Aouad, Learning from risks: a tool
for post-project risk assessment, Automation in construction 18 (1) (2008) 42–50. [17] I. Nonaka, The concept of‘Ba’: building a foundation for knowledge creation,
California Management Review 40 (3) (1998) 40–54.
[18] J.H.M. Tah, V. Carr, Knowledge-based approach to construction project risk management, Journal of Computing in Civil Engineering 15 (3) (2001) 170–177. [19] J. Raftery, Risk Analysis in Project Management, E&FN SPON, New York, 1994. [20] J.H. Woo, M.J. Clayton, R.E. Johnson, B.E. Flores and C. Ellis, Dynamic Knowledge
Map: reusing experts' tacit knowledge in the AEC industry. Automation in construction 13(2) (2004) 203–207.
[21] K.J. Yi, D. Langford, Scheduling-based risk estimation and safety planning for construction projects, Journal of construction engineering and management 132 (6) (2006) 626–635.
[22] K.R. Nielsen, Risk management: lessons from six continents, Journal of manage-ment in engineering 22 (2) (2006) 61–67.
[23] K.Y. Lin, L. Soibelman, Promoting transactions for A/E/C product information, Automation in Construction 15 (6) (2006) 746–757.
[24] M. Nissen, M. Kamel, K. Sengupta, Integrated analysis and design of knowledge systems and processes, Information Resources Management Journal 13 (1) (2000) 24–43. [25] M.W. McElroy, The New Knowledge Management: Complexity, Learning, and
Sustainable Innovation, KMCI Press/Butterworth–Heinemann, Burlington, 2003. [26] N.A. Kartam, S.A. Kartam, Risk and its management in the Kuwaiti construction
industry: a contractors' perspective, International journal of project management 19 (6) (2001) 325–335.
Table 8
Project RM performance measurement of thefield trials.
Project Project RM performance measurement
Time Cost Quality
A Project delivered on time No cost increasing Only very demanding applications are affected B Project delivered on time 2% cost increasing Only very demanding applications are affected C Project delivered on time Insignificant cost increasing Quality was acceptable
D b5% time increasing No cost increasing Only very demanding applications are affected E Project delivered on time Insignificant cost increasing Quality was acceptable
[27] O.A. Jannadi, S. Almishari, Risk assessment in construction, Journal of construction engineering and management 129 (5) (2003) 492–500.
[28] Organization for Economic Co-operation and Development, The Knowledge-based Economy, Paris, 1996.
[29] P. Carrillo, P. Chinowsky, Exploiting knowledge management: the engineering and construction perspective, Journal of Management in Engineering 22 (1) (2006) 2–10. [30] P. Demian, R. Fruchter, CoMem: evaluating interaction metaphors for knowledge reuse from a corporate memory, CIFE Technical Report, Stanford University, June 2004.
[31] P.F. Drucker, Post-capital Society, Harper Collins, New York, 1993.
[32] Project Management Institute ( PMI ), A guide to the project management body of knowledge (PMBOK guide), PMI Inc., Pennsylvania, 2004.
[33] R.V. Wyk, P. Bowen, A. Akintoye, Project risk management practice: the case of a South African utility company, International journal of project management 26 (2) (2008) 149–163.
[34] R.H. Clough, C.A. Sears, S.K. Sears, Construction Project Management, 4th edR, Wiley, New York, 2000.
[35] S. Baradan, M.A. Usmen, Comparative injury and fatality risk analysis of building trades, Journal of construction engineering and management 132 (5) (2006) 533–539.
[36] T.E. El-Diraby, B. Wang, E-society portal: integrating urban highway construction projects into the knowledge city, Journal of Construction Engineering and Management 131 (11) (2005) 1196–1211.
[37] T.E. El-Diraby, F. Briceno, Taxonomy for outside plant construction in telecommu-nication infrastructure: supporting knowledge-based virtual teaming, Journal of Infrastructure System 11 (2) (2005) 110–121.
[38] T.E. El-Diraby, C. Lima, B. Feis, Domain taxonomy for construction concepts: toward a formal ontology for construction knowledge, Journal of Computing in Civil Engineering 19 (4) (2005) 394–406.
[39] T.H. Davenport, L. Prusak, Working Knowledge, How Organizations Manage What They Know, Harvard Business School Press, Boston, 1998.
[40] T.H. Davenport, S.L. Jarvenpaa, M.C. Beers, Improving knowledge work process, Sloan Management Review 37 (4) (1996) 53–65.
[41] T. Lyons, M. Skitmore, Project risk management in the Queensland engineering construction industry: a survey, International journal of project management 22 (1) (2004) 51–61.
[42] T. Mezher, M.A. Abdul-Malak, I. Ghosn, M. Ajam, Knowledge management in mechanical and industrial engineering consulting: a case study, Journal of Management in Engineering 21 (3) (2005) 138–147.
[43] T.E. El-Diraby, J. Zhang, A semantic framework to support corporate memory management in building construction, Automation in Construction 15 (4) (2006) 504–521.
[44] T.E. El-Diraby, Web-services environment for collaborative management of product life-cycle costs, Journal of Construction Engineering and Management 132 (3) (2006) 300–313.
[45] T.M. Zayed, L.M. Chang, Prototype model for build-operate-transfer risk assess-ment, Journal of management in engineering 18 (1) (2002) 7–16.
[46] T.R. Gruber, Toward principles for the design of ontologies used for knowledge sharing, International Journal of Human–Computer Studies 43 (5–6) (1995) 907–928. [47] M.D. Tsai, H.P. Tserng, The application of knowledge management in Taiwan
con-struction industry, Proc., International Symposium on Automation and Robotics in Construction, IAARC, Jeju Island, Korea, 2004, pp. 183–187.
[48] U. Remus, S. Schub, A blueprint for the implementation of process-oriented knowledge management, Knowledge and Process Management 10 (4) (2003) 237–253. [49] U. Schultze, Investigating the Contradictions in Knowledge Management, Proc.,
IFIP '98 WG8.2 & WG8.6 Joint Working Conf. on Information Systems: Current Issues and Future Changes, vol. 9, IFIP, 1998, pp. 155–174, Helsinki, Finland, ISBN 3 901 882 02 2.
[50] W.C. Wang, J.J. Liu, S.C. Chou, Simulation-based safety evaluation model integrated with network schedule, Automation in construction 15 (3) (2006) 341–354. [51] Y. Mohamed, S. AbouRizk, Technical knowledge consolidation using theory of
inventive problem solving, Journal of Construction Engineering and Management 131 (9) (2004) 993–1001.
[52] Y. Rezgui, Ontology-centered knowledge management using information retrieval techniques, Journal of Computing in Civil Engineering 20 (4) (2006) 261–270. [53] Y.C. Lin, L.C. Wang, H.P. Tserng, Enhancing knowledge exchange through web
map-based knowledge management system in construction: lessons learned in Taiwan, Automation in Construction 15 (6) (2006) 693–705.
[54] Y.L. Yin, H.P. Tserng, M.D. Tsai, A model of integrating the cycle of construction knowledgeflows: lessons learned in Taiwan, Automation in Construction 17 (5) (2008) 536–549.
[55] T.R. Gruber, A translation approach to portable ontologies, Knowledge Acquisition 5 (2) (1993) 199–220.