Fuzzy mining is an adaptive process simplification technique based on multi-perspective metrics. This paper proposes a new approach to process mining that allows for different faithful simplified views of a process. The approach uses the metaphor of a roadmap to visualize process models, where activities and their relations are clustered or removed based on their importance. Traditional process mining techniques often produce "spaghetti-like" models that show all details without distinguishing what is important. The proposed approach uses significance and correlation metrics to determine which parts of the process model are important and which can be abstracted. The paper discusses the challenges of applying process mining to less-structured processes and introduces a framework for measuring significance and correlation. The approach is based on a configurable and extensible framework for measuring significance and correlation, which allows for different perspectives of the process. The paper also describes an adaptive graph simplification approach that removes edges and nodes based on their significance and correlation. The approach is implemented as the Fuzzy Miner plugin for the ProM framework and has been tested on various real-life logs. The results show that the Fuzzy Miner is able to clean up large amounts of confusing behavior and extract structure from chaotic data. The paper concludes that Fuzzy Mining is a first step towards more meaningful and applicable process mining techniques.Fuzzy mining is an adaptive process simplification technique based on multi-perspective metrics. This paper proposes a new approach to process mining that allows for different faithful simplified views of a process. The approach uses the metaphor of a roadmap to visualize process models, where activities and their relations are clustered or removed based on their importance. Traditional process mining techniques often produce "spaghetti-like" models that show all details without distinguishing what is important. The proposed approach uses significance and correlation metrics to determine which parts of the process model are important and which can be abstracted. The paper discusses the challenges of applying process mining to less-structured processes and introduces a framework for measuring significance and correlation. The approach is based on a configurable and extensible framework for measuring significance and correlation, which allows for different perspectives of the process. The paper also describes an adaptive graph simplification approach that removes edges and nodes based on their significance and correlation. The approach is implemented as the Fuzzy Miner plugin for the ProM framework and has been tested on various real-life logs. The results show that the Fuzzy Miner is able to clean up large amounts of confusing behavior and extract structure from chaotic data. The paper concludes that Fuzzy Mining is a first step towards more meaningful and applicable process mining techniques.