Enterprises define their business objectives in business processes, and workflows automate the business processes by completing tasks which realize parts of business goals in a particular order [1]. As a dominant factor in workflow management, developing appropriate analysis techniques for workflows is necessary [2]. Irani et al. state that workflow analysis facilitates locating problems in business processes and preventing repetition errors during workflow enactment. [3]. Vergidis et al. claim that workflow analysis adopts a range of different tactics such as simulation, diagnosis, validation, verification, and performance analysis to clarify the characteristics, potential conflicts, possible bottlenecks and any promising process alternatives [4].
To assure the correctness of workflow execution, analyses on structural integrity of workflows are widely studied. Adam’s methodology detects inconsistent dependencies among tasks to assure the safety of a workflow [5]. van der Aalst et al. develop effective Petri-net based techniques to verify deadlocks, livelocks (infinite loops), and dead tasks from workflow schemas [2][6][7]. In [8], Kiepuszewski et al. define structured workflow model and claim that a structured workflow is well-behaved, i.e. free from deadlock and multiple active instances of the same activity. Kiepuszewski et al. also claim that although structured workflow model is less expressive, most arbitrary well-behaved workflows can be transformed into a structured workflow, and structured workflow model is a good tool for various kinds of workflow analysis.
Besides, combining timing constraints into analysis of workflow models is also familiar.
Li et al. indicate that analysis of temporal factors is essential for validation of the interval
dependencies with temporal constraints in a workflow schema [9]. Adam et al. consider timing constraints as the external conditions for structural correctness of a Petri-net based workflow model [5]. Chen et al. develop an approach for dynamic verification of fixed-time constraints in grid workflow system [10]. From a graph based workflow model, Eder et al. develop a timed graph model to illustrate the working duration of activities among workflows with the corresponding earliest and latest finish time, and calculate the deadlines among internal activities to meet the overall temporal constraints on the basis of the model [11][12].
Marjanovic et al. build the timing model based on duration and instantiation space, and model the absolute and relative deadline constraints for dynamic verification [13]. Zhuge et al.
consider durations of activities for temporal checking in both design-time and run-time and model the temporal factors in workflows as timed workflow model for further analysis. [14]
For the organization perspective, modern WfMS regulates activities of employees through varieties of access control methods. Among the methods, role-based access control (RBAC) model [15][16] grouping users with similar permissions into roles is a popular solution among enterprises. However, business processes are operated based on not only roles but also tasks.
With both as core concepts, Oh et al. propose task-role-based access control (TRBAC) model to provide more modeling power for access control in WfMS [17]. Delegation which allows subjects like access rights or work items being authorized between users or roles during run-time is an interesting problem for workflow management [18] and is often studied on the basis of the corresponding access control model. For example, RBDM0 [19], RBDM1 [20], and the methods in [21], [22], and [23] describe various delegation models based on RBAC [15][16]. On the basis of RBAC, Crampton et al. describe an approach to manage delegation in WfMS, and raise several new issues about delegation of tasks for work-list-based WfMS [18].
Delegation for TRBAC is also studied in [24] and [25]. Jian et al. construct a framework and define the components for delegation in TRBAC [24], and Hsu et al. enhance the work by
considering temporal issues in [25].
A well-structured workflow may still fail or produce unanticipated run-time behavior because of abnormal data manipulation, the artifact anomalies. Detect artifact anomalies in workflows checks possible data misuse buried in workflow specifications. Various methodologies have been developed for detection of artifact anomalies generated from structural relationships between activities in a workflow [26][27][28][29][30]. Sadiq et al.
present seven basic data validation problems, redundant data, lost data, missing data, mismatched data, inconsistent data, misdirected data, and insufficient data in structured workflow model [26][31]. Hsu et al. define preliminary improper artifact usages anomalies, and introduce the analysis of such anomalies in design phase of a structured workflow [27][28]. In [29], Wang et al. introduce a behavior model to describe the data behavior in a workflow and refine the work accomplished in [28] by improving its efficiency. In [30], Hsu et al. raise the issues about analyzing artifact anomalies in workflows adopting message passing data models, and describe a formal description for such anomalies. Nevertheless, how temporal factors may affect the analysis of artifact anomalies is still seldom addressed. The methodology detecting artifact anomalies generated from twisted temporal and structural relationships between activities in workflows should be further discussed on the basis of the previous studies.
As for the resource perspective, Reveliotis et al. construct a Petri-based model with consideration of resource allocation, and uses the model for structural and deadlock analysis of workflow applications [32][33]. Based on Zhuge’s work [14], Li et al. estimate the active intervals of activities, and develops an algorithm to detect and remove resource conflicts with respect to both temporal and structural issues [34]. Zhong et al. adopt Li’s methodology [34]
onto a petri-net based workflow model, and develop an algorithm to detect resource conflicts when a new workflow being put into WfMS during run-time [35]. Based on [36], Hsu et al.
develop an incremental methodology for analysis of resource constraints in structured
workflows with temporal consideration during design-time [37]. The generation or elimination of resource conflicts are tracked and alerted along with each edit operation made by designers of workflows [37]. However, the technique for structural analysis adopted in [37] is inefficient and can be revised with the methods proposed in [30].
In this dissertation, structured workflow modeled in [8] is extended as temporal structured workflow (TS workflow) model with the temporal issues considered in [34] and [37]. The techniques for structural and temporal analysis on TS workflows are first introduced, and the methodology to analyze TS workflows in organization, data, and resource perspectives are then discussed. For the organization perspective, the works accomplished in [25] are refined to adopt TS workflow model for temporal constraints. A delegation framework for the WfMS coordinated with TRBAC model is established, and a series of algorithms for delegation of task instances and exploration of delegatees are developed. With the framework, a user is able to delegate his work to another user through an approval process, and WfMS can automatically delegate an emergent work item to an appropriate delegatee. The constraints such as elimination of delegation loops and separation of duty (SOD) are validated for delegation requested by either users or WfMS during run-time. As for the data perspective, a formal model describing artifact anomalies in TS workflow is established on the basis of define-use-kill operations. The issues about the artifact anomalies produced from twisted structural and temporal relationships between activities in a TS workflow are discussed and modeled. The methodology for static analysis of artifact anomalies buried in a TS workflow is developed.
Finally, for the resource perspective, the incremental methodology accomplished in [37] is refined to integrate TS workflow model and the analysis techniques proposed in [30]. The edit operations for constructing loop-reduced TS workflows (LRTS workflows) are first stated, and the methodology tracking down the generation and elimination of resource conflicts along with each edit operation made by designers is described.
The rest part of this dissertation is organized as following. In chapter 2, TS workflow model is sketched. The basic elements and the construction rules for TS workflow are described and the methods for analysis of temporal and structural properties in TS workflows are introduced. In chapter 3, a delegation framework for the WfMS coordinated with TRBAC model is introduced. In chapter 4, artifact operations and the corresponding artifact anomalies are first introduced, and the methodology detecting artifact anomalies in TS workflows is then described accordingly. In chapter 5, an incremental methodology tracking down the resource conflicts generated or eliminated in the steps of construction of an LRTS workflow is presented.
The related works for each of the above topics are discussed separately at the end of the chapters, and the conclusion and future works of this study are made in chapter 6.