Merging several business process variants

Abstract

In the last decade, modern sophisticated process-aware information systems have been taken place to provide a new possibility of process model configurations at build-time and enable process instance changes during runtime. However, this advantage has generated another challenge, which is the high price of configuration and maintenance of the big number of the derived process model variants (process variants for short). This paper proposes an algorithm that accepts as input a collection of process variants and generates a merged model. This algorithm has four main steps: determine the distinct blocks, common blocks, placeholders and finally construct the merged model. The merged model contains two types of blocks, common blocks and placeholders; the first block captures the commonalities of the process variants, and the second one captures the differences between them. In this way, the merged model is kept as small as possible. Furthermore, this merged model can subsumes the behaviors of all input models, ensures the trace back of each element from which input model is originated, and derives any of the input models from the merged model. Existing solutions either fail in respecting these requirements or allow only for merging pairs of process models. However, our algorithm allows for merging a collection of process variants at the same time.