Abstract. Continuously improved business processes are a central success factor for companies. Yet, existing data analytics do not fully exploit the data generated during process execution. Particularly, they miss prescriptive techniques to transform analysis results into improvement actions. In this paper, we present the data-mining-driven concept of recommendation-based business process optimization on top of a holistic process warehouse. It prescriptively generates action recommendations during process execution to avoid a predicted metric deviation. We discuss data mining techniques and data structures for real-time prediction and recommendation generation and present a proof of concept based on a prototypical implementation in manufacturing.

No abstract

No abstract

Abstract. The success of most of today's businesses is tied to the efficiency and effectiveness of their core processes. This importance has been recognized in research, leading to a wealth of sophisticated process optimization and analysis techniques. Their use in practice is, however, often limited as both the selection and the application of the appropriate techniques are challenging tasks. Hence, many techniques are not considered causing potentially significant opportunities of improvement not to be implemented. This paper proposes an approach to addressing this challenge using our deep Business Optimization Platform. By integrating a catalogue of formalized optimization techniques with data analysis and integration capabilities, it assists analysts both with the selection and the application of the most fitting optimization techniques for their specific situation. The paper presents both the concepts underlying this platform as well as its prototypical implementation.

Workflows are an established IT concept to achieve business goals in a reliable and robust manner. However, the dynamic nature of modern information systems, the upcoming Industry 4.0, and the Internet of Things increase the complexity of modeling robust workflows significantly as various kinds of situations, such as the failure of a production system, have to be considered explicitly. Consequently, modeling workflows in a situation-aware manner is a complex challenge that quickly results in big unmanageable workflow models. To overcome these issues, we present an approach that allows workflows to become situation-aware to automatically adapt their behavior according to the situation they are in. The approach is based on aggregated context information, which has been an important research topic in the last decade to capture information about an environment. We introduce a system that derives high-level situations from lower-level context and sensor information. A situation can be used by different situation-aware workflows to adapt to the current situation in their execution environment. SitOPT enables the detection of situations using different situation-recognition systems, exchange of information about detected situations, optimization of the situationrecognition, and runtime adaption and optimization of situationaware workflows based on the recognized situations.