Improving process models discovery using AXOR clustering algorithm

The goal of process mining is to discover process models from event logs. Real-life processes tend to be less structured and more flexible. Classical process mining algorithms face to unstructured processes, generate spaghetti-like process models which are hard to comprehend. One way to cope with these models consists to divide the log into clusters in order to analyze reduced sets of cases. In this paper, we propose a new clustering approach where cases are restricted to activity profiles. We evaluate the quality of the formed clusters using established fitness and comprehensibility metrics on the basis distance using logical XOR operator. throwing a significant real-life case study, we illustrate our approach, and we show its interest especially for flexible environments.