Herrmann, Thomas; Hoffmann, Marcel (1999): Augmenting Self-Controlled Work Allocation in Workflow-Management- Applications. In: Bullinger, Hans-Jörg; Ziegler, Jürgen (Eds.), Human-Computer Interaction: Communication, Cooperation and Application Design. Proceedings of HCI 99, Mahwah: Lawrence Erlbaum Associates, pp. 288-292. Augmenting Self-Controlled Work Allocation in Workflow-Management-Applications Thomas Herrmann, Marcel Hoffmann Informatics & Society, University of Dortmund 1. Introduction One of the main functions of workflow-management-systems (WMS) is to ensure that work is suitably allocated and executed on time. Current workflow technology supports work allocation through different concepts: Workflow- and role-models for assigning users to the steps of a business process. Allocation rules for directing work-items to users. Pools and agendas for offering work-items which have to be chosen by the users. Personal work-lists for summarizing already allocated work-items of an individual user. However, these concepts do not cover exceptional conditions, unforeseen properties of cases, and conflicting requirements that call for adaptations of prespecified work-allocation decisions. We assume that contradictory and continuously changing requirements of work allocation should not be solved by automatic rule evaluation and centralized planning approaches. From our point of view it is more advantageous to rely on the usersõ creativity and flexibility to optimize the work allocation system. We investigated requirements for flexibility of work allocation in case studies and developed concepts for self controlled work allocation with WMS. The case studies were part of the project MOVE (http://www.do.isst.fhg.de/move, 01 HB 9606) which is funded by the Federal Ministry of Education, Science, Research and Technology (BMBF). Section 2 introduces general objectives of work allocation, requirements from work-oriented design, and examples for flexible adaptation. Section 3 sums up our approach to augmenting self-controlled work allocation and the implications on technological and organizational design.
2. Objectives of work allocation methods 2.1. General objectives for the improvement of work allocation systems It is well known that allocation methods and decisions affect both organizational issues and the labor concerns. As a consequence, work allocation concepts within WMS must meet general objectives of business process improvement like reduction of time and costs or enhancement of quality and criteria for workoriented job design, like improvement of work conditions or flexibility. As WMS introduce a number of potentially negative effects it is of superior importance to evaluate and consider the criteria of job design during the design of work allocation methods. Literature provides many instruments to measure psychological criteria (e.g. Dunckel et al. 1993). Some Criteria that should be considered in the course of task allocation are: Variety and variability of tasks Task completeness Extension of temporal scope and reduction of time constraints Avoidance of backlogs 2.2. Requirements for flexible adaptation of allocation decisions in practice In many real life business processes unforeseen customer demands, errors, bottlenecks or any other kinds of exceptions call for adaptations of pre-specified allocation decisions. One reason for adaptation that we met several times in our case studies is escalation; other prominent ones are time constraints and workload balancing. Contract management for express services includes the agreement of discount rates between customers and the express service. When sales representatives and pricing analysts cannot agree on appropriate discount rates, cases are escalated to higher positions. Escalation of controversial cases is very difficult to plan in advance and must be supported with functionality for flexible delegation and additional information for informed decisions (see section 3.1. and 3.2.). When time becomes an important factor for signing a contract, clerks in the contract management take over activities from their colleagues. This valuable interpretation of roles can be supported by flexible role modeling and access to those pools which contain either preceding or subsequent work-items (3.3.). In the field of the forwarding business, workload-balancing is a critical success factor. For instance, the distribution of piece goods on vehicles for delivery has to fulfill many constraints such as maximum use of loading capacities, short road ways etc. Sometimes drivers participate in distribution decisions and negotiate with the distributor. Constraints and conflicting requirements can be controlled by negotiation, voting mechanisms and constraint based control (3.4.).
3. How to augment self-controlled work allocation? To support user-controlled work allocation, different views can be employed. For further information on the suggested concepts see (Hoffmann et al. 1999). 3.1. Flexible reservation, selection, rejection and delegation of work-items From the users point of view, technologically imposed restrictions in the course of selection, sequencing of tasks, rejection, and delegation must be reduced and the scope of action must be extended. Reservation: Users can annotate their preferences before they select a workitem. The WMS should support the implementation of different kinds of policies concerning the question of whether users have to respect the preferences of others or not. Selection: Users can select work-items from many different roles and corresponding pools. Presenting additional information in the agenda from which they choose facilitates the selection (3.2.). Flexible role- and workflow-models increase the scope for selection (3.3.). Inserting work-items while another one is pending: In principle, users should have the option of opening more than one window presenting work-items. However, this possibility should not be abused to make hidden reservations for the most interesting work-items. Rejection: Users might realize that they do not have the experience needed to carry out a work-item or that it is not possible to complete the work-item on time for personal reasons. In these cases or others, it should be possible to reject the work-item and to put it back on the work-list. Delegation: If users cannot complete a certain task and if they know the appropriate person to take care of this work-item, they should be able to delegate it to this person. However, the presupposition that the recipient of the delegated work-item has the right to carry out the task must be fulfilled. 3.2. Presenting additional context information Obviously, the workers can only employ the described possibilities for flexible selection of work-items if they have appropriate information as a basis for their decisions. Table 1 gives an overview of the most important items of information which should be shown in the agenda. Another type of information describes the situation of workers, e.g. the number of cases a person has carried out within a certain time period. Together with the information on work-items this information is continuously evaluated in order to automatically adapt allocation rules and to detect violations of constraints (3.4.)
Table 1: Context information for informed allocation decisions The date when the work-item has been submitted to the agenda The date by when the task has to be completed The roles (or persons) who are allowed to carry out the task Recommendation proposing a person who should carry out a work-item A hint about whether the work-item is partially completed Basic characteristics of the work-item (e.g. the value of an ordered product) Customer who is affected by the workitem Has the work-item been withdrawn or forwarded? Name(s) of person(s) who has already processed this case Hints concerning the completeness of data which is needed to process the case 3.3. Flexibility of workflow- and role-models Different types of bindings between users and tasks and vague modeling reduce imposed allocation decisions and increase the individual scope of action. Neighbor-roles assign preceding and subsequent tasks to userõs and thus provide access to preceding and subsequent activities. The temporary-role allows users to temporarily play different roles which are not permanently allocated to them. In contrast to conventional role playing relations, users are assigned to a larger set of temporary-roles which they may perform. Of course the selection of temporary-roles and the access to work-items from other usersõ agendas through neighbor-roles must be carefully controlled (3.4.). Vague modeling (Herrmann&Loser 1999) allows incomplete definitions of business process and uncertain assignment of tasks to users. A simple means for vague modeling is the role concept itself, which assigns tasks to groups of users without saying who exactly fills out occurring work-items. With late-modeling (Hagemeyer et al. 1997), parts of a workflow- and role model can be completed during the execution of a business process. 3.4. Controlling work allocation by negotiation, voting mechanisms and constraints Flexibility of workflow- and role-models and enhanced functionality for reservation, selection, etc. challenge the control of work allocation. How can deadlines and quality be ensured when many users concurrently and competitively perform allocation decisions? We suggest three concepts that complement each other. Negotiation: If user-controlled work allocation affects other users of the workflow-application, peers can solve conflicts by means of computer mediated negotiation among themselves (Herrmann 1995).
Voting mechanisms: To provide a fairer possibility for distributing workitems, it can be sensible to offer a voting mechanism. Highly developed voting concepts (Ephrati et al. 1994) record the usersõ reservations and selections and limit future selection and reservation abilities. Constraint based control continuously evaluates the performance of the entire workflow-application and alerts users when certain conditions or deadlines are likely to be violated. In some cases the WMS will also execute obligatory assignments. 4. Conclusion Means to support flexible work allocation are scarcely reflected in the literature on workflow management systems. The dominating principle of task allocation is based on role concepts where the management pre-specifies how the tasks of business processes are distributed amongst the staff. This concept is not suitable if the improvement of the employeesõ working conditions is an important objective. A large set of possibilities for improvement can be identified in the case of user-controlled work allocation. As we have shown, this can be achieved by extending the features of current systems. The next steps will be to present our proposals with the help of a prototype. This will enhance the participatory decisions in the companies who take part in our case studies. References Dunckel H. et al. (1993): Kontrastive Arbeitsanalyse im BŸro. Der KABA- Leitfaden. ZŸrich: Verlag der Fachvereine; Stuttgart: Teubner. Ephrati, E.; Zlotkin, G.; Rosenschein, J. (1994): Meet Your Destiny: A Nonmanipulable Meeting Scheduler. Proceedings of the CSCW Ô94: Transcending Boundaries, pp. 359-371. NY: acm-press. Hagemeyer, J. et al. (1997): FlexibilitŠt bei Workflow-Management-Systemen. Tagungsband Software-ErgonomieÕ97. Berlin: Springer. Herrmann, Th. (1995): Workflow Management Systems: Ensuring organizational Flexibility by Possibilities of Adaption and Negotiation. COOCS«95. Conference on Organizational Computing Systems, pp. 83-95. NY: acm-press. Herrmann, Th; Loser, K.-U. (1999): Vagueness in modes of socio-technical systems. Submitted to Behaviour and Information Technology, Taylor & Francis Hoffmann, M.; Lšffeler, Th.; Schmidt, Y. (1999): In. Herrmann, Th.; Scheer, A.-W.; Weber, H. (Eds.): Verbesserung von GeschŠftsprozesse mit flexiblen Workflow-Management-Systemen. (Band 3), to appear.