Interaction pattern detection in process oriented information systems

Finding interaction patterns is a challenging problem, but this kind of information about processes or social networks might be useful for an organization's management to understand the role of specific persons in processes. Ad-hoc processes are of special interest, because they result from runtime-collaboration between the participants, not using predefined models specifying the persons responsibilities and the order of activities. Because social network analysis (SNA) is closely related to interaction pattern detection, we introduce it as a method to determine properties of social networks like project teams. In order to support the detection of these patterns, we discuss the necessity of additional semantic activity information, and we propose rules and an algorithm that allow detecting such patterns automatically. We apply our algorithm in a case study, using Caramba to perform an example ad-hoc process. (C) 2006 Elsevier B.V. All rights reserved.