This paper describes the concept of workflow merge and methods for merging business processes. We grouped merges in four categories according to the type of merge: sequential, parallel, conditional, and iterative, and describe the corresponding algorithms for performing these operations. We give results that allow us to determine whether a merge operation is sound. It is shown that to avoid invalid merges, one should choose merge points between which a sub-workflow, called a merge region, is well structured. These findings can provide useful guidance for future workflow merge research. We also raise issues of more complex merge problems, such as merge conflicts, semantic ambiguities and workflow splits.
For agile business operation, modern corporations must make frequent business process changes as well as organizational changes through mergers and acquisitions. In 2001, Hewlett-Packard Company and Compaq Computer Corporation announced a merger agreement to create an 87 billion dollar global technology leader. The merged company offers the most complete set of products and services in the IT industry with expected cost savings of approximately 2.5 billion dollar a year [10]. Many important issues arise in integrating the two giant organizations, one of them being how to integrate their business processes. Since frequent changes in business processes and operations are becoming increasingly common, both through internal reorganizations and through mergers and acquisitions, we have conducted preliminary research on how workflows can be modified dynamically to adapt to such changes. In addition, our research provides support for complex process composition, i.e., creating complex workflow processes from simple ones.
A workflow may be modified at a schema level that defines a workflow process or at an instance level that represents a specific instance of an already defined process [12]. For example, a two-step workflow, “place order,” then “deliver,” represents a simple business process at the schema level. Within this process there may be orders or instances of a process “order” for customer “Sue”. In this paper, we define the workflow concept and methods focused at the schema level. This research topic is important and has not been fully addressed by previous research efforts. Process reengineering and evolution have been studied from different perspectives such as: business process integration [5] and [17]; using generic workflow modeling methods to ensure flexibility [2]; describing methods for workflow evolution [12]; improving workflow interoperability [3], [9], [11] and [15]; and enhancing exception handling capabilities [14] and [19]. Dealing with more than one process makes a workflow merge different from other problems that commonly assume a single process, and makes existing methods inefficient for addressing workflow merging issues. Therefore, this topic does not fit well into the existing research frameworks. For example, the classic process integration methods [5] and [17] collaboratively bridge, adapt, and exchange information without actually modifying the processes of the business partners. Naturally, when companies merge, both process integration and merging of processes are necessary for streamlining their operations.
In the remainder of the paper, we first introduce a workflow modeling method with Petri nets. Next, in Section 3, we introduce the workflow merge concept and validate our approach. Section 4 introduces notions of sound and unsound merges, and gives two results related to soundness. Then, in Section 5, we discuss merge point detection and other issues such as conflicts, semantic ambiguities, and impact of merges on organizational roles and resources. Finally, Section 6 gives brief concluding remarks.
This paper discussed fundamental concepts, models, and methods of various types of workflow merge operations. We formally defined a workflow merge operation and also proposed merge methods. Merges were grouped into four categories: sequential, parallel, conditional, and iterative. More importantly, we showed the conditions under which a merge will yield a sound result. This framework is promising for developing applications to serve the business world, and its potential benefits lie in: (1) performing simulations that can help decision makers visualize merges for business processes; (2) creating virtual enterprises that make flexible business operations possible; and (3) planning merges that allow software agents to share process knowledge. It also offers a systematic approach for building complex workflows from simple ones by incorporating changes and new sub-processes into them in a correct way. Thus, this methodology also plays a useful role in workflow evolution.
Decision makers can gain valuable experience through simulating their proposed workflow merge operation. The simulation results can provide crucial information to help them modify their plans and optimize the new processes. Such an effort will eventually help to cut costs, increase throughput, and create more value.
The research on workflow merge also promises more flexible business models, such as virtual enterprises, to address dynamic business environments. A virtual enterprise is a business model that dynamically organizes small companies into its business processes. A virtual enterprise can provide more services in a flexible manner and lead to more efficiencies as compared to a single enterprise providing multiple services. Moreover, such coalitions can disband when they are no longer effective [16]. At present, coalition formation for virtual organizations is limited. We anticipate that automation of coalition formation by using workflow merge technologies will save both time and labor. In complex settings, workflow merge techniques may also be more effective at finding better coalitions than other technologies or manual methods.
In a multi-agent environment, if a software agent's plan (essentially a process) is incomplete, other agents may be able to help with the other parts of the solution. Then the problem is how to assimilate such piecemeal solutions into the existing processes, which is equivalent to merging several plans [8]. As workflows and plans are both based on process specifications and can use similar representations, we believe that workflow merge technologies can be applied in plan merging too.
In summary, research on workflow merge offers a new way to connect business processes. This paper has highlighted the importance of the workflow merge problem and also raised many questions, such as resolution of merge conflicts, semantic considerations, workflow splits, etc., for future studies.
