Discovery and Analysis of Activity Pattern Co-Occurrences in Business Process Models

Abstract

Abstract:Research on workflow activity patterns recently emerged in order to increase the reuse of recurring business functions (e.g., notification, approval, and decision). One important aspect is to identify pattern cooccurrences and to utilize respective information for creating modeling recommendations regarding the most suited activity patterns to be combined with an already used one. Activity patterns as well as their cooccurrences can be identified through the analysis of process models rather than event logs. Related to this problem, this paper proposes a method for discovering and analyzing activity pattern co-occurrences in business process models. Our results are used for developing a BPM tool which fosters the modeling of business processes based on the reuse of activity patterns. Our tool includes an inference engine which considers the patterns co-occurrences to give design time recommendations for pattern usage.

Crowd-Based Mining of Reusable Process Model Patterns

“…We were not able to do this in time. However, the works by Lau et al [8] and Greco et al [3] are very close to our M achine algorithm: they share the same underlying frequent sub-graph mining technique. We therefore expect a very similar performance.…”

“…We also intend to adapt the Crowd algorithm to BPMN, to compare it with other BPMNoriented approaches in literature [8,3], and to study if the crowd can also be used for quality assessment (to automate the complete pattern mining process).…”

“…Lau et al [8] propose to use frequent sub-graph and association rules discovery algorithms to discover frequent sub-graphs (patterns) to provide modeling recommendations. Li et.…”

Automatic Generation of Optimized Process Models from Declarative Specifications

“…The paths of existing process models are extracted and analyzed with association rule mining. [13] [15] extend an existing process model, while our approach generates one from a declarative specification. [9] requires annotations of the existing process models.…”

“…[13] does not allow building a process model with an and-Split and therefore is not sufficient in our case. [15] predicts which activity pattern (generic process fragment) will follow the partly modeled process. The paths of existing process models are extracted and analyzed with association rule mining.…”