Aggregating Individual Models of Decision-Making Processes

When faced with a difficult decision, it would be nice to have access to a model that shows the essence of what others did in the same situation. Such a model should show what, and in which sequence, needs to be done so that alternatives can be correctly determined and criterions can be carefully considered. To make it trustworthy, the model should be mined from a large number of previous instances of similar decisions. Our decision-process mining framework aims to capture, in logs, the processes of large numbers of individuals and extract meaningful models from those logs. This paper shows how individual decision data models can be aggregated into a single model and how less frequent behavior can be removed from the aggregated model. We also argue that main process mining algorithms perform poorly on decision