Flexible Process-Aware Information Systems De ciency Management in Construction

Abstract

De ciency management (DM) is an important sub eld of the construction domain which is characterized by a high demand for immediate and exible reactions to unexpected problems. Thus, there is a high potential for exible process-aware information systems. We propose a de ciency management system (DMS) to support the DM process in a exible manner supported by knowledge-sharing of bestpractice processes. We acquired a set of requirements concerning process support and knowledge sharing for DMS and present rst steps towards the development of a working prototype.

Introduction

To retain the competitive advantage of today's companies, the streamlining of business processes is increasingly important to develop new performanceenhancing features, to accelerate the internal e ciency, and to reduce costs [ 1,17 ]. Moreover, the economic success of a company heavily depends on its ability to exibly respond to changes in its environment and to take advantage from arising opportunities. Hence, the ability to quickly change processes or to deviate from a pre-set course of action is essential. As a consequence processaware information systems (PAISs) are a desirable technology in many domains as these systems support the operational business of a company based on models of the organisation and its processes [17]. PAISs include traditional work ow management (WFM) systems as well as modern business process management (BPM) systems. Current research particularly addresses approaches for increasing the exibility of PAIS [ 1,4 ]. Recent research in PAIS has also recognized the need for knowledge management through process reuse from best-practice process collected in repositories [ 1 ]. Knowledge sharing and reuse becomes a central prerequisite for enabling process exibility, in order to address the increased need for decision making on the process level.

De ciency Management (DM) is an important sub eld of construction domain that particularly needs to deal with unforeseen changes, demanding high exibility by all involved parties. Generally speaking, a de ciency in construction is a negative deviation of the actual state of construction of a building from the speci ed or expected conditions [15, p. 5]. Thus, a de ciency is always unexpected and requires immediate remedial actions, which lead to changes of the current plans. Therefore, we expect that there is a high potential for exible PAIS in the eld of DM [ 7 ]. In this paper, we derive a set of requirements concerning process exibility and process reuse in de ciency management systems (DMS) that are relevant for future more advanced approaches. We present a related concept as well as rst steps towards the development of a working prototype. 2

Flexibility in Process-Aware Information Systems

A PAIS is \a software system that manages and executes operational processes involving people, applications, and/or information sources on the basis of process models" [ 4 ]. In order to operationalize process models, a PAIS typically includes a WFMS as a generic component for the execution of work ows. Traditional WFMSs strictly separate build time and run time of a work ow. During build time, a work ow de nition is created to operationalize a business process (or a part of it). During run time, this work ow de nition is repeatedly instantiated to execute the occurring business cases in exactly the same manner over and over again. For about 15 years, various approaches are discussed to address the exibility needs of PAIS [ 5,16,13,17 ]. Schonenberg et al. [17] present a classi cation of exibility approaches into four types: exibility by de nition, exibility by change, exibility by deviation, and exibility by under-speci cation. Flexibility by de nition refers to the ability to consider alternative execution paths in the process model during the modelling (build time). Traditional WFMS already support this type of exibility. It can only take into account foreseen and predictable events and changes. Flexibility by change describes approaches that permit changes of process de nitions and/or instances during run-time while maintaining consistency. Flexibility by underspeci cation refers to the ability to execute process descriptions which are not fully speci ed. Thus, certain decisions can be deferred to an appropriate point in time during process execution. Late modelling and late binding are two techniques used for this purpose. Flexibility by deviation refers to the forth and so far only rarely explored class of approaches that o ers the possibility that the real-world process execution di ers from the modelled process without the need to modify the process de nition in advance.

An essential characteristic of all exibility approaches is the fact that process modelling and execution are not strictly separated any more as in classical work ow systems [ 1,13 ]. Thus, a modi cation or a late modelling during runtime can be considered a re-modelling of the work ow that immediately e ects its execution. However, modifying a work ow requires signi cant skills in the domain as well as in process modelling. Decisions must be taken concering how the work ow is modi ed and how this modi cation is formalized in the underlying work ow modelling language. Hence, methods are required that support users in performing such modi cations. ADEPT/AristaFlow [16] and CAKE [ 3,13 ] are two advanced work ow systems which support exibility by change and underspeci cation and which include methods to support users in reusing best-practice work ows. 3

Introducing Process Flexibility into De ciency

Management Systems

Based on an analysis of current DMS in construction available on the German market in summer 2014 [ 6 ] and four interviews with construction experts, we now derive a set of requirements for future more advanced DMS. We mainly focus on those requirements that related to the support by exible PAIS. Additionally, we show how these requirements can be implemented based on the generic framework CAKE for integrated process and knowledge management. To provide a general understanding of the DM, we sketch the main processes and characteristics of this domain rstly. 3.1

De ciency Management in Constructions

The entire DM process consists of several sub-processes addressing di erent steps of the overall DM process. It begins with logging a de ciency. General information like issuer name and company are recorded, as well as speci c information such as de ciency description, oor, space, required action, and additional issuer notes. Motzko and Racky [ 14 ] point out the importance of a comprehensive record keeping of the de ciency in a centralized and dedicated area due to regulatory demands and further business needs. The entered information is veri ed and complemented by a responsible person inside the construction company. Next, the particular kind of de ciency must be identi ed. For example, de ciencies like wall cracks are described by di erent criteria such as appearance (e.g. a single or a bundle of cracks), crack width, or possible impacts such as impairment of the structural integrity. Depending on the reported de ciency, a visual on-site inspection is necessary in order to de ne the correct de ciency type and to decide an applicable recti cation method. Further, it has to be check if it is eligible for de ciency recti cation (DR) under warranty, for instances, if the de ciency was caused by liable a subcontractor. In addition the DR has to be assigned to construction workers for execution. The processing of the DR has to be controlled until its completion to ensure the timely nalization in accordance with the contract. Therefore, the accurate tracking of all open DRs is important. Overall, DM must incorporate all involved parties, including all subcontractors [ 14 ]. DMS aims at supporting DM by providing IT support for the relevant activities involved in DM. Today, a large variety of DMS exists, with di erent strengths and weaknesses. Thus, we analysed the process exibility approaches implemented by these systems and determined that more advanced exibility approaches than exibility by de nition are not implemented in any of the investigated systems [ 6 ]. 3.2

User Requirement

The below described requirements have been derived based on the results of the interviews, also considering general regulatory requirements in DM [ 8 ] where necessary. We illustrate the most important requirements by examples, providing typical use cases of an envisioned future DMS.

R1: Support for process exibility. Future DMS must enable exibility by change and by underspeci cation. In particular, ad-hoc changes of work ow instances must be supported as well as late binding. An additional argument in favour of these methods is the fact that process instances must remain aligned with the activities in the real world in order to support tracing of DM processes as well as their accurate documentation. Further, exibility by de nition is required due to the high process variability. Example: An assigned construction worker receives a work order to x a wall crack based on a process for elastic sealing. On the construction site s/he investigates the cracks and after cleaning s/he recognises that there is no need for an elastic sealing as due to drying shrinkage it is likely that the cracks will stabilize. Therefore, a structural strengthening non-elastic injection system is more e ective. Using a mobile device, the worker immediately modi es the running process. S/he deletes those activities that are not relevant any more, i.e., \Prepare elastic sealing material" and \Apply two layers with the injection system". Then, s/he adds the activity \Prepare nonelastic injection material" followed by the activity \Inject material into crack" at the appropriate position in the work ow instance.

R2: Collaboration support and role-based access control. A processoriented collaboration platform is needed for coordinating all activities involved in DM and for supporting the necessary communication and documentation needs. As a large variety of parties and persons are involved in DM, such a platform requires a role-based access control. The access control must enable a detailed control of access rights for all resources, in particular on the level of individual tasks. Example: The investigation of a reported de ciency yields that it was caused by a liable subcontractor. Hence, several tasks of the DR process must be assigned to some of the subcontractor's employees. Therefore, the responsible project leader from the subcontractor accesses the DMS. Due to her/his access rights s/he is able to make the respective assignments, but only for those tasks, his company is in charge of.

R3: Knowledge-sharing of best-practise processes. Future DMS sup

porting process exibility should also support knowledge-sharing of best-practice work ows. Successfully nished DM processes should be captured and stored in a repository. Further, the reuse of best-practice work ows should be supported, thus asking for appropriate means for navigation and search in the repository. This requirement particularly arose in the interviews and is considered a means to improve e ciency and quality in the context of the large variability in DM. Example: After the successful termination of the DR process for repairing a crack, the project leader can store the particular work ow instance as a bestpractise work ow in a repository. During this process, all case-speci c data is removed and the work ow is generalised towards a work ow de nition. At some later point in time, a similar type of crack must be xed. By search in the repository, the previously stored work ow de nition is found. It can be instantiated (and adapted if necessary).

In addition to these three requirements addressing process exibility and closely related aspects, several further, more general requirements must be met as well by future DMS. For example, the usability of such a system must be particularly ensured by an intuitive user interface enabling to control the exibility functions of the PAIS. Also a simple graphical modelling language for work ows is required to enable sta from construction companies to perform work ow modelling without the need to involve their IT personnel. Further, providing access to the DMS via mobile devices is important, as deviations from planned DR processes are mostly discovered at the construction site. 3.3

CAKE { Collaborative Agile Knowledge Engine

We now brie y describe CAKE, a generic framework for integrated process- and knowledge management [ 3 ] and explain how it could be used to ful l the identi ed requirements. CAKE integrates an agile work ow engine with a so-called knowledge engine that supports process reuse as a particular kind of knowledgesharing. The agile work ow engine is used for the enactment of agile work ows and supports their collaborative modelling and adaptation in a consistent manner. The work ow engine enables exibility by change as it allows users to model and change work ow de nition and instances at any time, provided that the user is granted the respective access right. Further, a simple graphical modelling language is available that allows modelling, execution monitoring, and adaptation of work ows within a browser-based editor. The modelling language includes placeholder task, thus late binding is supported. So, CAKE is in line with requirement R1.

The purpose of the knowledge engine is to support users in nding, de ning, and adapting work ows according to their current needs. The knowledge engine implements a process-oriented case-based reasoning (CBR) method [ 12 ]. CBR is an established AI methodology for problem solving based on the assumption that similar problems have similar solutions [ 2 ]. The CAKE knowledge engine maintains a repository of work ows which can be semantically annotated using terms from a domain ontology. It supports work ow reuse by similarity-based retrieval of work ows. Thus CAKE is in line with requirement R3.

Finally, the CAKE framework consists of a storage layer that implements a role-based access control mechanism for all resources managed by CAKE, in particular work ows, tasks, documents, services, etc. The access control mechanism is a decentralized discretionary access control with subject-object relationships speci ed in access control lists. Thus CAKE is in line with requirement R2. The overall CAKE software is implemented as Web-based system. The client user interfaces enable access to all work ow related functions such as work ow modelling, execution, similarity-based retrieval, and adaptation using a standard browser. Further, the CAKE Server API also allows mobile applications to directly connect to CAKE, e.g., to support the mobile execution of tasks on an Android-based device. 3.4

Applying CAKE for Building a Prototype for De ciency

Management

We build the rst version of a prototype for demonstrating and evaluating the bene ts of process exibility in DM. For this purpose, we selected a sub eld of frequently occurring de ciencies in construction, namely cracks in facades and masonry1. We collected technical background documentation and a set of process descriptions of respective DR processes from the construction companies contacted during the interviews. Based on this documentation, we developed an ontology of the relevant DM tasks and building materials, as well as a de ciency ontology to classify and describe di erent kinds of cracks. Further, we formalized the provided process descriptions using the CAKE work ow editor and thereby we created a repository of initial best-practice work ows. The resulting domainspeci c CAKE instance can then be used to support the DR process as illustrated in the following use case.

Use Case: A series of cracks is a reported to the project manager of a construction company. An initial assessment of the cracks takes place, leading to a description of the cracks w.r.t. the de ciency ontology. Based on this description, the project manager searches for applicable work ows in the best-practice repository. S/he selects a DR work ow for \elastic sealing of cracks", starts a new work ow instance for this case, and assigns a construction worker to the repair tasks. While performing the rst activity of the work ow, which is the cleaning of the cracks, the worker recognises that most of the cracks exceed the maximum size allowed for sealing with a exible injection system. To clarify which alternative method could be used, s/he searches for similar work ows in the best-practice repository. The knowledge engine of CAKE retrieves several work ows which are similar to the present work ow but which are in addition suitable for larger cracks (see Fig. 1). After inspecting the proposed work ow s/he decides to apply the method of stitching instead of sealing for repairing the crack. Thus, the work ow editor of CAKE is used to adapt the work ow to include the new activities \drill holes on both sides of the crack" and \grout in ushaped metal units". The worker now follows the adapted work ow to complete the repair, guided by the CAKE work ow engine. 4

Conclusion, Related, and Future Work

In this paper, we have revealed that DM in construction is a new and very promising application area for exible PAIS. Due to the fact that de ciencies 1 See http://theconstructor.org/concrete/methods-of-crack-repair/886/. always occur unplanned and often require an immediate remedial action, it is essential that resulting process changes are performed during runtime. High process exibility ask for exibility approaches by change and/or underspeci cation. We have presented a concept and an initial prototype for a PAIS that addresses these needs based on the prototypical generic software system CAKE for integrated process and knowledge management and brie y demonstrate its feasibility by a rst use case.

The application of process management and work ow systems in the construction industry is discussed for more than ten years in the scienti c literature. Ruppel and Klauer [ 9 ] developed a work ow application that supports construction projects, but they do not consider the needs of process exibility related to foreseen and/or unforeseen changes. Korten [ 11 ] comprehensively investigates the use of information technology to support construction processes. The BauVoGrid project aims at developing a grid-based framework for supporting construction processes. It also includes work ow components as well as semantic technologies. The project results demonstrated in the eld of DM [ 10 ]. However, process exibility is not addressed by this project. Although the investigation of unforeseen changes within process execution in construction was already considered in 2005 [9, p. 118], we are not aware of any in-depth research on this topic so far.

Future work will focus on tailoring CAKE for DM and on developing a full prototype that can be demonstrated to the construction industry. Based on such a prototype, more detailed case studies concerning the usability and potential bene ts in DM can be performed. Another potential direction of future research is the investigation of new approaches for exibility by deviation, which have the potential to void the need for explicit work ow adaptation. Such approaches might be able to better support processes in DM where only the purpose is known in advance, but not the precise order of steps that need to be executed. 15. Oswald, R., Abel, R.: Leitfaden uber hinzunehmende Unregelmassigkeiten bei

Neubauten. Bau-Verlag (1996) 16. Reichert, M., Weber, B.: Enabling Flexibility in Process-Aware Information Systems: Challenges, Methods, Technologies. Springer, Berlin-Heidelberg (2012) 17. Schonenberg, H., Mans, R., Russell, N., Mulyar, N., Aalst, W.v.d.: Process Flexibility: A Survey of Contemporary Approaches. In: Advances in Enterprise Engineering I, pp. 16{30. No. 10 in LNBIP, Springer Berlin Heidelberg (2008)

References

1. Van der Aalst , W.: Business process management: A comprehensive survey . ISRN Software Engineering ( 2013 )

2. Bergmann , R.: Experience Management: Foundations , Development Methodology , and Internet-Based Applications , LNAI, vol. 2432 . Springer ( 2002 )

3. Bergmann , R. , Gessinger , S. , Gorg, S. , Muller, G.: The Collaborative Agile Knowledge Engine CAKE . In: Proceedings of the 18th International Conference on Supporting Group Work . pp. 281 { 284 . GROUP '14, ACM , New York, NY, USA ( 2014 )

4. Dumas , M., van der Aalst , W., ter Hofstede , A.: Process-aware Information Systems: Bridging People and Software Through Process Technology . John Wiley & Sons, Inc., New York, NY, USA ( 2005 )

5. Fleischmann , A. , Schmidt , W. , Stary , C. , Augl , M. : Agiles Prozessmanagement mittels Subjektorientierung . HMD Praxis der Wirtschaftsinformatik 50 ( 2 ), 64 { 76 ( 2013 )

6. Frei , K. : Mangelmanagement im Bauwesen { Marktubersicht von Software zum Mangelmanagement und Nutzenpotenziale agiler Prozessunterstutzung . Master thesis , University of Trier ( 2014 )

7. Gessinger , S. , Bergmann , R.: Potentialanalyse des prozessorientierten Wissensmanagement fur die Baubranche . In: Henrich, A. , Sperker , H.C. (eds.) LWA 2013 . Lernen, Wissen & Adaptivitat ; Workshop Proceedings. pp. 212 { 219 ( 2013 )

8. Hechtl , A. : Kooperation im Bauwesen . Bautechnik - Zeitschrift fur den gesamten Ingenierbau 79 ( 6 ), 396 { 401 ( 2002 )

9. Klauer , T. : Eine prozessorientierte Kooperationsplattform fur Bauprojekte auf Basis eines internetbasierten Work ow-Managements . Ph.D thesis , Shaker ( 2005 )

10. Klose , S. , Angeli , R. , Dollmann , T. , Ernst , T. , Fellmann , M. , Hilbert , F. , Hoheisel , A. : Spezi kation der mobilen Work ow Enactment Engine, der entwickelten Referenzprozessmodelle und der mobilen Dienste . Tech. Rep. BauVOGrid-Bericht A5.1 ( 2010 )

11. Kortgen, M.: Optimierungsansatze zur prozessorientierten Abwicklung komplexer Bauma nahmen unter Einsatz neuer Informations- und Kommunikationssysteme . Ph.D thesis , Kassel University Press, Kassel ( 2010 )

12. Minor , M. , Montani , S. , Recio-Garc a , J.A.: Process-oriented case-based reasoning . Information Systems 40 , 103 { 105 ( 2014 )

13. Minor , M. , Tartakovski , A. , Schmalen , D. , Bergmann , R.: Agile Work ow Technology for Long-Term Processes - Enhanced by Case-Based Change Reuse . In: Methodological Advancements in Intelligent Information Technologies: Evolutionary Trends , pp. 279 { 298 . IGI Global ( 2010 )

14. Motzko , C. , Racky , P. : Erfolgsfaktoren fur eine reibungslose Abnahme . Baumarkt und Bauwirtschaft 101 , 30 { 33 ( 2002 )