A workflow model is useful for business process analysis. A well-built workflow can help a company streamline its internal processes by reducing overhead. The results of workflow modeling need to be managed as information assets in a systematic fashion. Reusing these results is likely to enhance the quality of the modeling. Therefore, this paper proposes a document-based workflow modeling mechanism, which employs a case-based reasoning (CBR) technique for the effective reuse of design outputs. A repository is proposed to support this CBR process. A real-life case is illustrated to demonstrate the usefulness of our approach.
Reengineering has enabled companies to operate faster and more effectively and to use information technology more productively (Hammer & Stanton, 1999). Workflow models have become popular because of such increased interests in business process reengineering and improvement of related technologies such as inter-networking and object-oriented techniques (Jablonski & Bussler, 1996). They help managers fight their ways through the productivity wars of the past 10 years. Companies have revamped their processes and cut out redundant activities. Workflow technologies can facilitate enterprises' requests for improved competitiveness by increasing productivity and enhancing services for customers (Lawrence, 1997).
Workflow models are useful for business process analysis (Bracchi and Pernici, 1984 and Ellis and Nutt, 1980). When workflow technology is adopted in real business, it is important to model workflow in a systematic fashion. Very few modeling experts encounter the same case more than once in their entire career; modeling experience is valuable and hard to acquire (Brown & Gupta, 1994). The results of modeling activities are valuable information assets. It is important to manage these modeling results systematically.
Case-based reasoning (CBR) technique is useful for the reuse of knowledge. CBR is a problem solving technique that reuses past cases, experiences, or tacit knowledge (Kim & Han, 2001b; Kolodner, 1991; Kolodner, 1993, Korczak et al., 1989, Noh et al., 2000, Riesbeck and Schank, 1989 and Slade, 1991). CBR retrieves similar cases from a case base, selects the most similar case among them, adapts this case according to the user's requirements, and then stores this new case into the case base for future reuse.
Although CBR is useful for the reuse, it has not been applied for workflow modeling. However, CBR has been employed successfully for other modeling activities such as data modeling (Lee and Han, 1997 and Paek et al., 1996). Therefore, this paper proposes a document-based workflow modeling support system (DWMSS) using the CBR technique. DWMSS is better able to utilize the business know-how embedded in the previous workflow cases.
In Section 2, we introduce and sharpen a document-based workflow model (Lee & Suh, 2001). Section 3 presents a real-life case that is modeled by the use of DWMSS. Section 4 explores the system architecture and modeling process of DWMSS. Section 5 describes a repository included in DWMSS. 6 and 7 illustrates the details of the modeling process of DWMSS by the use of a real-life case. Section 8 compares DWMSS with other modeling support systems using the CBR technique. Section 9 concludes this paper and points out some important avenues for future research.
Workflow models are useful for not only helping employees understand the whole picture of their job, but also streamlining the business processes they share with others; the accumulated workflow models are valuable knowledge assets of a company. Therefore, it is important to manage them in a systematic fashion for future reuse. For this purpose, this paper proposes a workflow modeling system by employing a well known case-based reasoning technique. This system can helps companies or consulting firms fight their ways through the productivity of workflow management. The proposed repository is better able to help reuse the valuable business process know-how embedded in previous workflow models.
In spite of its practical usefulness, DWMSS can still be improved. The followings are further research avenues. First, DWMSS still lacks the capability of understanding semantics used in names for workflows, tasks, or documents. It may require mechanisms beyond similar term indexes. In addition, enhancing the maintenance of our case base by the use of genetic algorithms may be of interest (Kim & Han, 2001a). Second, automatic learning mechanisms on the basis of the current cases may be of further help. Lastly, the current DWMSS needs to improve users' interface.
