Data Retrieving Agent Test

Data retrieving agents provide external data-retrieving functions that integrate external data into the DBPCS so that data distributed through GC and A/E information systems can be shared via the DBPCS with all actors. This test was designed to verify data-retrieving activities soundness (see Figure 6.17).

In the test case, three data-retrieving agents were developed to retrieve, respectively,

“D13. Subcontractor information table”, “D14. Subcontractor evaluation table” and “D15.

Qualified subcontractor list” data classes for the collaborative design process. Figure 56.44 shows these three data retrieving agents running in the DBPCS.

Three running Data Retrieving Agents in the DBPCS

Figure 6.44 Three Data Retrieving Agents Running in the DPBCS

We took as our test example the D13 data-retrieving agent of the “D13. Subcontractor information table”. When an actor wants to submit subcontractor data(m) to the Assistant Agent Client Program data class “D13. Subcontractor information table” that he/she has generated in the external information system (see Figure 6.45), the D13 data retrieving agent will prompt for input query keywords (see Figure 6.46) required to retrieve subcontractor data(m) from the external database. Figure 6.47 shows the XML file, including retrieved data(m), which replicates the external data shown in Figure 6.45. This verifies the soundness of the D13 data retrieving agent. Employing the same methodology, we verified the soundness of D14 and the D15 data-retrieving agents as well.

Subcontractor Data(m) in the External System

Figure 6.45 Example of Data(m) in the External Information System

Query Keyword Input Window

Figure 6.46 Example of Query Keyword Input to the Data Retrieving Agent

The retrieved Data(m) form the external system

Figure 6.47 XML file Containing the Retrieved Data(m) form the External System in Figure 6.39

Chapter 7

Conclusion and Recommendation

This chapter presents research objectives, summarizes findings and assesses to what extent objectives were met. Research contributions and lessons learned from IDBF experiments are also addressed. In conclusion, this chapter discusses recommendations as to future prospects for the IDBF.

7.1 Review of Objectives

The primary objective of this study was to facilitate fast-tracking construction in D/B projects through A/E and GC process collaboration within the D/B project team. The IDBF framework was subsequently developed in the paper to describe a collaborative process-based fast-tracking mechanism capable of achieving fast-track construction for a D/B project. Three sub-objectives achieved to facilitate IDBF implementation included (1) development of a fast-tracking model creation method based on the AD technique necessary to creating a fast-tracking model; (2) creation of the COPReM to generate the A/E and GC collaborative process model within the D/B team, and (3) development of the DBPCS as the IBDF infrastructure within which created fast-tracking and collaborative process models are implemented.

7.2 Summary

The Integrated Design-Build Framework (IDBF) was proposed as the medium through which fast-tracking construction in a D/B project would be implemented. Three models must be incorporated in the IDBF, including (1) the fast-tracking model, (2) the collaborative process model, and (3) the design-build process collaboration system

(DBPCS).

Apply AD methodology to create the Fast-Tracking Model:

In line with fast-tracking philosophy, this study applied the AD method to divide a design-build project into several DBMs with overlapping relationships. Each DBM, composed of a collaborative design process and collaborative construction process, was responsible to accomplish a specific construction component. Therefore, unlike the traditional linear design and construction mechanism, DBM overlapping has the potential to shorten a D/B project delivery time.

Using semantic similarities to generate the Collaborative Process Model:

For cross-organization process collaboration within D/B project teams, this study focused on the crucial stages of semantic similarity analysis and process re-engineering method to develop the Cross-Organization Process Re-engineering Method (COPReM), employed to identify the collaborative linkages between original A/E and GC processes.

The COPReM comprises the four stages of (1) process representation, (2) semantic hierarchy creation, (3) semantic similarity analysis and (4) cross-organization process reengineering. Reasonable collaborative process models for connecting A/E and GC’s processes can be schemed out through a systemic approach.

Semantic similarity analysis, the core of the COPReM, allows for process comparisons to be done easily; the results of which can then be used to evaluate process similarities. This study applied the following five process-related similarity measures: (1) name affinity, (2) name set affinity, (3) process information similarity (ISim), (4) activity similarity (ASim) and (5) function similarity (FSim). Based on these measures, overlapping and coupling analyses were applied to determine the process duplications and

information flows between A/E and GC processes so that the collaborative process model could be generated using process/activity functional mergence and data flow creation methods.

To validate COPReM feasibility, this research applied the model to an actual residential building project case study. Analysis results not only demonstrated the appropriateness of applying semantic similarity analysis to process integration, but also validated COPReM feasibility.

Develop the Design-Build Process Collaboration System:

The DBPCS was developed based on Microsoft BizTalk architecture and the multi-agent system philosophy. The three primary components incorporated within the DBPCS were the (1) DBM management module; (2) process engine; and (3) activity execution module. The DBM management module managed and controlled all DBMs in the imported fast-tracking model.

In the DBM management module, the execution status of each DBM can be updated through cooperation between the DBM manager and process monitor. Once updated, the DBM manager can implement the proper DBM in the process engine based on the updated status and overlapping relationships of DBMs in the fast-tracking model.

Once the DBM management module implements DBMs, the process engine (developed on a Microsoft BizTalk server) can drive the imported collaborative process model. The activity execution module, developed based on multi-agent system architecture, is subsequently responsible to delegate invoked activities to the proper actors automatically in accordance with collaborative process model activity information. DBPCS actors can then fulfill delegated activities with the assistance of agents. In other words, actors

responsible to accomplish a specific activity can collaborate with other members via data retrieval, data sharing, and message communications, and tasking to achieve common activity goals. Accordingly, the DBPCS provides a platform to not only integrate the distributed efforts of actors within A/E and GC organizations, but also connect actors together to accomplish activities that share a common objective.

7.3 Conclusions

The IDBF provides a process-collaboration-based mechanism to facilitate fast-track construction in D/B projects. Relevant analysis methods were developed to facilitate the creation of the IDBF and a semi-workflow system, i.e., the DBPCS, was developed as the IDBF infrastructure. Facilitated by the DBPCS, D/B team actors can cooperate with one another to accomplish DBM collaborative processes necessary to implement the fast-tracking model. Conclusions based on research observations and experience follow below:

The Fast-Tracking Model can be created by decomposing a D/B project into DBMs.

A D/B project can be divided into DBMs by zigzagging decomposition between customer needs (CNs), function requirements (FRs), design proposals (DPs) and construction process variables (CPVs), with the design matrix [A], construction matrix [B]

and concurrent matrix [C] (derived from the zigzagging decomposition) describing the design and construction sequences between each DBM. Overlapping relationships between DBMs can be identified based on sequence information within the matrices. Thus, a fast-tracking model can be created by summarizing the DBMs and overlapping relationships.

The business process similarity can be evaluated by semantic similarity analysis.

This research focuses on process collaboration within a D/B project team.

Determining commonalities between two processes is key to identifying areas of potential collaboration between A/E and GC processes. Because a process aggregates activities and data entities, activity and data entity similarities represent proper indices of commonalities between two processes. Therefore, this study applies semantic similarity analysis to evaluate process similarity.

To calculate process similarity, all data entities and activity names in the analyzed processes must be clustered into, respectively, data and activity hierarchies, which describe the semantic distance between either two data entities or two activity names., Affinity functions are developed based on semantic distance to calculate data similarities and activity name similarities. The function similarity (FSim), information similarity (ISim) and activity similarity (ASim) of two processes can also be calculated to identify commonalities between two processes.

Using process overlapping and coupling analyses can discover the cross-organizational collaborations between two processes.

Duplicated activities and potential information flows must be determined in order to create the CPM. As activities duplicated in two processes have similar functions and data entities, they offer the potential to be merged when the two processes are required to cooperate. All merged activity resources, e.g., manpower, documents and efforts, must be integrated to maximize resource usage. Potential information flows include the data linkages of the two processes which can keep data consistency within the processes.

This research developed process overlapping and coupling analyses in order to, respectively, identify duplicated activities and potential information flows. Process overlapping analysis uses similarities related to process function, information and the

activities to determine which parts are duplicated in the two processes. Process coupling analysis investigates data affinities of the two activity’s input and output data to determine potential information flows. The A/E and GC collaborative process model can be created by summarizing the overlapping and coupling analysis results.

The created Fast-Tracking Model can be automated in the DBPCS based on overlapping relationships.

Overlapping relationships, as defined in the fast-tracking model, are programmed as rules in the DBPCS. Therefore, as the created fast-tracking model is imported into the DBPCS, the system can execute the model automatically based on overlapping relationships between DBMs.

The CPM can be automated using workflow techniques.

Using the Microsoft BizTalk Server, the CPM can be mapped to an “Orchestration”, a workflow model executed automatically in the DBPCS process engine. As the CPM is executed, each activity will be invoked and delegated to actors. Therefore, the DBPCS provides a semi-workflow mechanism with which to implement the CPM.

Distributed architects and engineers of A/E and GC companies can be integrated through the multi-agent system architecture.

This study applies the multi-agent system to assist actors to execute CPM activities.

As the process engine invokes an activity, a corresponding activity agent is created to manage autonomously the execution of that invoked activity. Firstly, the activity agent can delegate the activity to those actors responsible for its completion. Actors can then handle an activity following their original protocol with the assistance of an assistant agent responsible to feed the actor with requisite data and information. After actors finish a

certain number of activity tasks, the assistant agent reports all relevant achievements to the DBPCS in order that the progress of various actors can be integrated and shared with other actors. Meanwhile, actors can communicate with other activity members via the activity agent. Thus, activity teamwork collaboration can be realized through the cooperation of agents within the DBPCS.

Using the XML schema definition and software agent techniques, data distributed across different organizations or locations can be integrated into the DBPCS.

Because activity actors are distributed over A/E and GC companies, activity output data generated by different actors are naturally distributed over disparate data storage and information systems. To integrate distributed output data, data retrieving and management agents were developed to retrieve specific data from external databases and other means of digital storage. In this study, two data types are retrievable by agents. The first, “Digital Files”, include data stored in file systems (e.g., text document, image, CAD files) that are retrievable by uniform resource location (URL). The second type, “Database Records”, include data stored in databases that are retrievable by data retrieving agents installed on the DBPCS. Each retrieving agent is composed of a database adapter and data description.

The data adapter provides a network connection tunnel to the specific external database.

The data description is an XML schema definition file that identifies query parameters necessary to retrieve data from the database and requisite retrieved data attributes.

Therefore, using different database adapters and data descriptions, data retrieving agents may be developed to acquire data located in A/E and GC external databases.

The IDBF can facilitate fast-track construction through process collaboration.

In the IDBF, a D/B project can be decomposed to DBMs. Each DBM can be fulfilled by the collaborative process model generated with the cross-organization process

re-engineering method. Although processes embedded in DBMs may increase the execution complexity of the fast-tracking model, this study applies workflow and multi-agent system techniques to develop the DBPCS as the infrastructure for fast-tracking model execution. Thus, the DBPCS can implement and manage the fast-tracking model with collaborative processes. In summary, the IDBF provides a process collaboration-based mechanism to facilitate fast-track construction in D/B projects.

7.4 Research Contributions

The contributions of this research include:

(1) This is the first research to apply a process reengineering mechanism to facilitate fast-track construction.

(2) This research pioneers the application of cross-organization process re-engineering in D/B project teams and the promotion of process collaboration technologies in the construction industry.

(3) This research applied AD methodology successfully to identify fast-tracking DBMs and the overlapping relationships between them. It proposes a new fast-tracking analysis method for the design-builder to create a practical fast-tracking model.

(4) This research redefines original DBM overlapping relationships in the fast-tracking model. Six overlapping sequence types are defined to accurately describe the fast-tracking between DBMs.

(5) This research successfully evaluates process similarities along several vectors, including function similarity (FSim), information similarity (ISim) and activity similarity (ASim), based on semantic similarity philosophy. Process similarities

identify the commonalities within two processes, thus permitting further analysis of interactions between processes based on identified commonalities.

(6) This research developed two process collaboration analysis methods, overlapping and coupling, designed to identify duplicated activities and potential new information flows linking A/E and GC processes. Cross-organizational integration potential can be determined based on identified overlapping and coupling relationships to create the CPM of two organizations.

(7) This research proposed the CPM structure in order to describe the potential for cross-organizational collaboration between A/E and GC processes. Following the model structure, process managers can easily depict the two organization’s activity interactions and information flows. Moreover, since the CPM was developed based on A/E and GC original processes, A/E and GC companies can continue performing the CPM without changing their current processes.

(8) This research combined the workflow and multi-agent system technologies to create the new DBPCS semi-workflow system to integrate staff and activity efforts distributed throughout A/E and GC organizations. DBPCS system architecture can be the prototype for a practical construction industry workflow system.

7.5 Discussions

Implementation of the Created Fast-Tracking Model

Resource trade-offs is the critical issue in fast-tracking model implementation. In the fast-tracking model creation phase, matrices [A] and [B], respectively, represent the relationships between FRs and DPs and between DPs and CPVs. Often, analysts can

evaluate [A] and [B] based only on qualitative reasoning (i.e., the laws of nature) and available facts (e.g., design/construction schedule sequences, constraints, regulations).

Resource trade-off conditions are improper and difficult to include because of the complexity they introduce to analyses. Hence, the created fast-tracking model presents a sequence plan without resource allocations. Consequently, overlapping design or construction activities may be modified by schedule planners as sequential tasks due to resource trade-off considerations.

Implementation of the Created Collaborative Process Model

Although, this study proposed a method by which to reengineer cross-organizational processes within a D/B team and analyzed a residential building project case study to validate the feasibility of COPReM, the feasibility of the created collaborative process model was not validated because the model was not actually applied to the D/B project team. In order to ensure the feasibility of the proposed collaborative process model, the following pre- and post-COPReM implementation issues must be discussed:

Pre-COPReM Implementation Issues: The requisite collaborative process models for a D/B team should be identified prior to implementing COPReM. “What collaborative process models are necessary for this D/B team?” is an essential first question requiring an answer. The answer can hopefully be found in resolutions and/or responsibility clauses in the D/B alliance agreement. Objectives and scopes for collaborative process models being created then must be defined based on the answer to the question above and identify where overlaps between the two original entities is considered undesirable. Moreover, as different collaborative process models have their specific corresponding existing processes, e.g., design professional and GC, input processes properties, such as objectives, scopes, and primary data items, must be defined precisely prior to analysis.

After collaborative process models have been created, a further modification in accordance with the D/B team alliance agreement may be necessary. Although overlapping and coupling relationships can be depicted within the collaborative process model based on process information and function features, additional relationships may still need to be established based on D/B alliance agreement clauses. Accordingly, the process managers can append newly created relationships to the created models to make them practicable.

7.6 Recommendations

Three recommendations are suggested for future research on this subject.

Scheduling technology is essential to implement the fast-tracking model.

Few scheduling methods have been developed for fast-tracking construction.

Although the overlapping relationships identified in this research provide sequential information on fast-tracking phases that can be applied to the development of a fast-tracking network, resource allocation conditions remain a significant unknown and, while complex, need to be evaluated. Difficulties faced in estimating the actual reduction of construction duration may decrease the feasibility and practicability of fast-tracking construction. Therefore, in order to address this issue, the introduction of appropriate scheduling technology, which incorporates resource leveling and resource allocation methodologies, is critical for model success. Therefore, as a next step, an advanced fast-tracking scheduling method that considers resource trade-off conditions should be developed for the fast-tracking model creation method proposed in this paper .

Defining a common uniform semantic data thesaurus will be a critical element in achieving cross-organizational integration in the construction industry.

A data thesaurus provides a common linguistic framework essential to sharing and integrating information between two independent units (e.g., databases, information systems, processes). It gathers and provides the semantic relationships and the corresponding affinity values between data distributed over different units. The data thesaurus provides the common framework that allows the examination / assessment of relationships and similarities between data items; an activity critical to interpreting the semantic meanings of information stored on different systems and facilitating communication / interaction across system platforms. The data thesaurus is, therefore, essential to further system integration in the construction industry. As creating a uniform data thesaurus will demand a significant investment in time and other resources, a suggestion would be to use the data dictionary of the construction industry as a starting point and reference for development. This already prepared reference gathers together relevant construction industry terminology and assigns corresponding data attributes.

A workflow system will be critical to process integration

To integrate cross-organizational processes, an automatic data exchange mechanism between processes will be important. However, the low process automation degrees of A/E and GC decrease the possibility of implementing automatic data exchange between processes. To overcome this problem, this study created the DBPCS to provide a semi-workflow infrastructure for implementing a collaborative process model for A/Es and GCs. Although data-retrieving agents can integrate data distributed over A/E and GC companies into the DBPCS, actors still need to inform the system to retrieve specific data

To integrate cross-organizational processes, an automatic data exchange mechanism between processes will be important. However, the low process automation degrees of A/E and GC decrease the possibility of implementing automatic data exchange between processes. To overcome this problem, this study created the DBPCS to provide a semi-workflow infrastructure for implementing a collaborative process model for A/Es and GCs. Although data-retrieving agents can integrate data distributed over A/E and GC companies into the DBPCS, actors still need to inform the system to retrieve specific data