Understanding Spaghetti Models with Sequence Clustering for ProM

The goal of process mining is to discover process models from event logs. However, for processes that are not well structured and have a lot of diverse behavior, existing process mining techniques generate highly complex models that are often difficult to understand; these are called spaghetti models. One way to try to understand these models is to divide the log into clusters in order to analyze reduced sets of cases. However, the amount of noise and ad-hoc behavior present in real-world logs still poses a problem, as this type of behavior interferes with the clustering and complicates the models of the generated clusters, affecting the discovery of patterns. In this paper we present an approach that aims at overcoming these difficulties by extracting only the useful data and presenting it in an understandable manner. The solution has been implemented in ProM and is divided in two stages: preprocessing and sequence clustering. We illustrate the approach in a case study where it becomes possible to identify behavioral patterns even in the presence of very diverse and confusing behavior.