مدلسازی فرآیندهای کسب و کار با سیستم های جریان کار: ارزیابی دیدگاههای جایگزینی

Effective business process management necessitates a consistent information flow between the participants in the process, the smooth integration of the flow of work, the timely sharing of data and information during the planning and implementation phases and harmonious support for the collaborative aspects of work. The recent trends in the development of advanced workflow management systems and technologies seem to be of crucial importance for facilitating these tasks within the process management context. However, workflow management systems (WfMS) follow various approaches in modelling the flow of work and hence present varying functionalities when supporting enterprise processes. The present paper examines the ways in which workflow technology may facilitate the implementation of process management, reviews the pros and cons of adopting alternative workflow representation techniques in modelling business processes and provides guidance to managers as to the characteristics, the similarities and differences of the various workflow modelling schemes.

Today's dynamic business environment is driving a new extended organisation, which competes globally focusing on low prices and customer customisation of products and services. As a key success factor for effective competing one could identify the management of core business processes, which deliver value to their customers, suppliers and internal staff. Thus by focusing on automating, optimising, and continuously improving the core business processes, organisations can make commitments to those customers, employees, partners, and suppliers establishing a solid competitive advantage. Since the 1980s, Information Technology has provided a wide range of applications supporting automation and management of the business process. Workflow management systems (WfMS) are the most evolved of those applications providing consistent information flow between the participants in the process, smooth integration of the flow of work, timely sharing of data and information during the planning and implementation phases and harmonious support for the collaborative aspects of work. However, WfMS follow various approaches in modelling the flow of work and hence present varying functionalities when supporting enterprise processes. The implications of these approaches to the real-world management of processes and projects are not always clear and transparent to managers. As the different techniques match more or less to different types of processes, managers have to identify which approach to adopt. The objectives of the present paper are: to examine the ways that workflow technology may facilitate the implementation of business processes; to review the pros and cons of adopting alternative workflow modelling techniques in modelling the processes; and to provide guidance to managers as to the similarities and differences of the various workflow modelling schemes and their characteristics. This paper is organised in the following manner. Section 2 of the paper reviews current approaches in WfMS, while Section 3 outlines the alternative workflow modelling techniques. Section 4 illustrates a comparison of applying the two major workflow modelling techniques in a case-study. The case refers to the project management of a multi-annual European Union (EU) Operational Programme for Greece. Finally, Section 5 provides the conclusions and lists issues to be taken into consideration by managers when examining the adoption and use of one or the other workflow modelling system.

The analysis so far has shown that the two approaches usually applied in workflow management systems, i.e. the activity- and the communication-based, can provide adequate vehicles for process modelling when the aim is an effort to reengineer and/or streamline these processes aiming for operational improvement. None of the two approaches falls behind the other concerning the scope and wealth of information provided. Specifically, they both provide: • ample information concerning the tasks to be modelled (separating them into either phases, or tasks and activities), • rich information for the time duration of processes and the cost incurred for the execution of the processes, and • constructs for modelling issues such as parallelism and if–then–else mechanisms. However, the two approaches present a number of differences concerning the “philosophy” of modelling which should be taken explicitly into account and are strongly related to the limitations that, in general, the workflow systems have. Below we list the basic ones, observed throughout the modelling procedure, which constitute the main scope of the paper: • the methodological rigour implied by each approach, i.e. the possibility of formally specifying (using provable mathematical constructs) the correctness of each approach. At that point, we can clearly differentiate the two approaches given the different philosophies underlying them. The communication-based approach is implicitly using an underlying model (the one described thoroughly by Winograd and Flores, 1987), but the activity-based one seems to lack a strong mathematical construct. However, this observation relates to a general limitation of workflow modelling: the lack of rigorous modelling. Such efforts to present workflows analytically took place recently, using either meta-graphs or distributed computational modelling techniques (see Weske, 1998;Basu & Blanning, 1998). The concluding remark is that obviously the activity-based approach can be more easily formalised in the context of these two analytical essays, since they both are activity-oriented analytical techniques. • The simplicity/complexity in applying the approaches, especially with regard to the flow of communication and the interdependence of tasks. Both methods are highly dependent on the type of the process. The key issue, though, is that the activity-based approach enables us to observe possible interdependencies of tasks but does not provide much information about the flow of communication among processing entities (Weske, 1998). On the contrary, that is being brought to an end successfully by the communication-based approach, which, in turn, due to its structure, cannot represent a large number of interdependent tasks. • The ease of applying the approaches to well defined repetitive processes. It is very hard, for different reasons in each approach, to model exceptional tasks or processes. Therefore, such tasks should be excluded, due to the uncertainty either in time or in the processing entities involved. • The ability and ease of each approach for handling client-orientation. The activity-based approach is more convenient for automating internal tasks of organisation, while the communication-based one strongly focuses on the relationship with the customer, reminding at each single step who the client is and what he is expecting from the task. • The managerial implications concerning the expertise required by each approach. The activity-based techniques seem to be less structured and hence require some process modelling expertise by the project management team. On the other hand, the communication-based technique is more structured and guides to an adequate degree (at least conceptually) the model designer throughout the whole modelling procedure. However, the latter can also be a disadvantage, because it restricts the degrees of freedom that the design team has. Such a case does not exist in the activity-based approach, which enables the modelling of even complex models. In addition to the above-mentioned issues, two remarks about the use of workflow modelling point to possible future research directions. First, the workflow representation techniques do not include uncertainty handling in their modelling. We should recall that in the process implementation, uncertainty drives big delays; see e.g. Adler, Mandelbaum, Nguyen, and Schwerer (1995). Time duration is usually represented in a deterministic way, without enabling or involving possible knowledge about the uncertainty. It would be very interesting to incorporate such probabilistic views which would couple very well with the “if–then–else” properties of the workflow systems, directing the flow of information to the appropriate processes. Second, the resource allocation mechanisms are not sufficiently covered by workflow representation techniques. Of course, the assigning of roles to different processing entities is, somehow, a resource allocation procedure. However, there is no clear representation of a resource allocation map as in most management software tools. Further extending and improving the resource handling issues of workflow modelling techniques could facilitate their adoption and use in business processes automation.