Object-aware process management

Abstract

Process management systems (PrMS) offer promising perspectives for a more flexible and efficient process execution. However, for many process-aware application systems, the underlying process logic is still hard-coded. As a consequence, these business applications are both complex to design and costly to maintain. One reason for this situation stems from the fact that contemporary PrMS were primarily designed for the support of highly structured, repetitive business processes. By contrast, many processes found in practice are rather unstructured or semi-structured, i.e., they are knowledge-intensive and driven by user decisions. This thesis first analyzes real processes not adequately supported by existing PrMS and elaborates their characteristic properties. As a major insight it became clear that in many application scenarios comprehensive process support requires both object- and process-awareness. To comply with the underlying data structure, processes must be based on two levels of granularity: object behavior and object interactions. Further, the individual processes, coordinating the behavior of single object instances, must be coordinated with the ones of related object instances. Opposed to these well-defined process support granularity levels, activities must be executable at different levels of granularity; e.g., a user may want to process a number of related object instances in one go. Furthermore, process execution must be accomplished in a data-driven manner. Finally, authorized users must be able to access and manage process-related objects at any point in time. The major contribution of the thesis is the PHILharmonicFlows framework, which enables object-aware process management in a comprehensive manner. The framework not only provides a new process modeling approach, but also establishes a well-defined operational semantics enabling the automatic and dynamic generation of end-user components at run-time.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 26 citations exist upstream.