In January of 1992, the Collaboration Technology Research Group (CTRG) was formed by Professors Clarence Ellis and Gary Nutt to research topics of human collaboration and computer support for cooperative work. Ellis had just joined the University of Color ado Computer Science Department, and this signalled the reunion of Nutt and Ellis who had previously worked closely together at Xerox PARC in the 1970s in the "Office Information Systems" area. In the year before CTRG was formed, Ellis was chief architect of the Bull Corporation workflow product, and Nutt worked with students at Colorado to create the Olympus/Escalante modeling system for parallel process modeling [6].
CTRG works on the principle that the state of the art can be advanced by close coupling of theoretical and applied research. The research group is concerned with theories, models, architectures, implementations, and studies / evaluations of computer suppo rted group activity. This is an inherently interdisciplinary topic, so CTRG draws together faculty and students from the Computer Science Department, the Business School, and the Institute for Cognitive Science at the University of Colorado.
Contemporary organizations employ a vast array of computing technology to support their information processing needs. There are many successful computing tools designed as personal computing aids (word processors, spreadsheets, etc.) but fewer tools desig ned for collaborating groups of people. These latter tools are called groupware . Groupware is defined as "systems that support groups engaged in a common task or goal, and that provide an interface to a shared environment." [1] Potential benefits and pitfalls of groupware have been discussed in conferences, tutorials, and journals concerned with the new discipline of Computer Supported Cooperative Work (CSCW.) [4] As opposed to much of the previous generation of office information systems literature, this literature addresses the inherently interdisciplinary nature of grou pware. To successfully implement groupware in an organization requires well grounded technological development, with careful attention paid to the social and organizational environment into which the technology is being imbedded. CTRG has a number of ongo ing research projects exploring groupware implementations and environments.
Many groupware products associated with the world wide web have recently been introduced; it is a hot topic. A few of these products capture knowledge of the organizational activity that they are assisting; but the vast majority do not. For example, a gro up document editor knows nothing about the organizational purpose of the document being edited. An electronic mail system knows not why the messages are being sent. Organizationally aware groupware can potentially lead to significantly more powerful and u seful systems. One class of organizationally aware groupware is workflow .
In 1994, CTRG received funding from the Information Technology and Organizations program (ITO) within the U.S. National Science Foundation (NSF) to research workflow systems. Workflow systems are designed to assist groups of people to carry out work proce dures, and contain organizational knowledge of where work flows in the default cases. Workflow is defined as "systems that help an organization to specify, execute, monitor, and coordinate the flow of work items within a distributed office environmen t." [5] The system typically contains two components. The first component is the workflow modeling subsystem, which assists administrators and analysts to define procedures and activities and their orderings and attributes. It allows analysis, simula tion, assignment of activities to users, etc. This component is frequently associated with business process reengineering. The second component is the workflow enactment subsystem consisting of the execution interface as seen by the end users, and the exe cution environment which assists in coordinating and performing the procedures and activities. It enables the units of work to flow from one users' workstation to another as the steps of a procedure are completed. Some of these steps may be executed in pa rallel, some may be executed automatically by the computer system. Our research explores the hypothesis that the modeling and enactment components need to be tightly interwoven, and can be useful to all participants at all phases of the workflow life cycl e.
Workflow systems and techniques have existed for decades but, despite progress in many areas, intelligent adequate workflow systems are still not well established. The models themselves tend to be too restrictive, and the enactment systems lack flexibilit y, scalability, distribution, and a solid theoretic foundation. The CTRG workflow research is addressing these issues. To advance the state of the technology, the research has the two objectives of:
(1) creation of a framework, and a family of mathematical models of organizational activity and workflow;
(2) development of system architectures that enable flexible, evolutionary large scale workflows.
Our research has endeavored to go beyond this classical definition of workflow to cope with issues arising from the nature of this work: it is sometimes highly creative, unstructured, situated, and dynamic. We have built a workflow framework based upon go als and dynamic functionality, rather than procedures and tasks. One PhD thesis worked upon within CTRG explored the definitions and dimensions of exceptions within workflow systems. Another thesis in progress is using Petri net models to explore dynamic structural change. Dynamic in this case means that a process may be reengineered without stopping, aborting or quiescing the systems. Jobs and transactions in progress, can continue in the midst of the structural change.
These are both examples of significant issues that require creative conceptualization, and thoughtful research to make significant progress. To help deal with issues such as these, we advocate explicit meta-models (to create, view and manipulate models), and corresponding meta-systems (to create, view, and manipulate workflow systems). A recent publication describes our conceptual approach which focuses upon multiple dimensions of workflow meta-systems [2]. We have also implemented a research prototype wo rkflow system on the internet, called Chautauqua, which illustrates our philosophy and our design strategy [3]. Chautauqua implements novel features to address exception handling and dynamic structural change. The implementation and usage of this system has provided a go od vehicle for testing of our concepts and implementation ideas. It has also been a catalyst for ideas of other useful groupware tools and features.
Workflow systems have also been strongly criticized as being coercive and dictatorial, because they impose a manager's view (or the management consultant's) of the work process upon everybody. This formalization is frequently NOT the way that the work act ually gets done. The work patterns can be very idiosyncratic, ad-hoc, and creative. We believe in the concept of subservient workflow. The system represents one view of the work processes, and only gives a suggestion of the way that work can be viewed and performed. It is not coercive, because users are free to ignore this statement of how work was done in the past, and in Chautauqua, we have implemented an experiment in end user workflow systems evolution. For more information on Chautauqua, visit the ho me page of CTRG [7].
There are many examples illustrating the importance of creative human activity within organizational and group work. Many organizations have large volumes of procedures manuals which are typically not read by anyone. It has been noted that an organization thrives on the problem solving and creative capabilities of people. Important information is transmitted by informal means such as discussion over lunch or over the coffee machine. If these people want to destroy an organization, they only need to blindl y follow the procedures manuals exactly and precisely. Ultimately, we see a need for the incorporation of adequate social and organizational sub-models and sub-systems into workflow, and generally into collaboration models and systems. This re quires a deeper understanding of group functioning, cognition, and personality. A marriage of social science and computer science is necessary to investigate some of the fundamental issues in this area which are currently poorly understood, and largely ig nored by contemporary groupware vendors.
Ellis, C. A., S. J. Gibbs, and G. L. Rein, "Groupware: Some Issues and Experiences," Communications of the ACM, Vol. 34, No. 1 (January, 1991a), pp. 38-58.
Ellis, C. and Nutt, G. "Multi-Dimensional Workflow" in Proceedings of the IDPT 96 Conference, December, 1996.
Ellis, C. and C. Maltzahn "Chautauqua: A Flexible Workflow System" Proceedings of the 30th HICSS Conference, January, 1997.
Grudin, J. "Why CSCW Applications Fail," Proceedings of the CSCW88 Conference, ACM, pp. 85 - 93.
Jablonski, S. and C. Bussler, Workflow Management Systems: Modeling, Architecture, and Implementation, Thomson Press, 1996.
McWhirter, J. D., DG: A Framework and System for the Specification and Instantiation of Graph Models, University of Colorado Computer Science Dept. Ph.D. Dissertation, 1993.
Home page of Collaboration Technology Research Group (CTRG): http://www.cs.colorado.edu/~skip/ctrg.html