Fuzzy Workflows-Enhancing Workflow Management with Vagueness

Workflow Management Systems have proven their positive effects in well-structured procedures. In practice business processes depend on implicit knowledge and require decisions based on unclear objectives. This vagueness of real-world problems isn’ t appropriately supported by modelling methods and workflow systems so far. Today the fuzzy set theory is used successfully in systems that provide sophisticated control mechanisms with a small set of simple rules. This concept hasn’ t been adopted yet within the services sector. In this paper a holistic and integrated advance for the integration of the fuzzy set theory into BPM from modelling methods to workflow automation is proposed. 1. Vagueness in business process automation Workflow management systems have positive effects on timeand cost-effectiveness in well-structured procedures. Especially standardized workflows that are repeated very often and which have a simple composition are suited for automation systems. Thus comprehensive software products are available that can support thousands of transactions a day e. g. the processing of customer orders in retail enterprises or the granting of loans in a bank. This business process automation is very often preceded by a business process reengineering project to identify appropriate procedures and to improve their quality and efficiency. Therefore the introduction of a workflow management system must be embedded into a holistic business process management concept from requirements analysis via business process modelling and the conceptual design to implementation and maintenance.[1] A lot of methods have been developed in order to model procedures from various perspectives.[2] Only a few of them are generally accepted. One example for structured methods are PetriNets[3] that are used very often in the context of workflow control[4] whilst Event-driven Process Chains (EPC) [5] are a semi-structured modelling method used on the conceptual level. These different fields of application result from the intuitive usage of EPCs for employees in contrast to the unambiguity of Petri-Nets that corresponds to the clearness that computers need as an input. In practice business processes depend on implicit knowledge and require decisions based on unclear objectives.[6;7] Decisions based on vague or qualitative information belong to the class of the decisions under uncertainty extending the classical view of deterministic and stochastic models.[8] In this contribution vagueness is understood as the uncertainty regarding data and their interdependences. A more detailed view on this term can be delivered by the identification of different kinds of vagueness referring to their origin.[9;10] The complexity of the environment and the perception limits of human beings cause informational vagueness. Business processes contain information from various sources or data with a short life-span and therefore at a fixed time only one part of the whole system can be analysed, so that data become obsolete during the collection of other partial aspects. Human preference profiles are not determinable. Thus the objective "substantial reduction of the processing time" cannot directly be transferred into actions as the extent of the desired change is unclear. Furthermore interdependences with other goals are not specified. Natural language descriptions of real world facts contain inherent (also: linguistic) fuzziness. Both the creation of linguistic models and the context sensitivity of linguistic statements contribute to the emergence of this vagueness. Closely connected is the inaccuracy in linguistic comparisons. The statement "this object value is much higher than x" can serve as example for this vagueness. Similarly it is a human characteristic to categorize perceived facts. E. g. a customer order is classified as high, medium and low. The boundaries cannot be determined exactly. These complex scenarios aren’ t appropriately supported by IT so far. This weakness derives from the difficulties to translate the vagueness and ambiguity of natural language and human thinking into business models and formal mathematical models for workflow automation. So far imprecise or tentative data have been treated as hard facts and sophisticated interdependencies have been reduced to a few relations as substitutes. Thus the transformation of real-world uncertainty, imprecision, and vagueness into information technology controls that are based on propositional logic imply a loss of relevant information. The use of vague data is favourable whenever appropriate measuring methods are missing or are too expensive, the environment is characterised by high dynamics or is not determinable. The whole business process management approach must be able to treat vagueness and fuzziness as shown in figure 1. Real-World Problem Vagueness, Imprecision, Uncertrainty. Crisp Modeling Crisp Business Process Models Crisp Translation Crisp Worlkflow Control / System Fuzzy Business Process Models Fuzzy Translation Fuzzy Worlkflow Control / System F uzzy M oeling F uzzy E xtnsion F uzzy E xtnsion Figure 1. Fuzzy BPM approach