Abstracting processes into a hierarchical structure of subprocesses helps to improve the understandability and readability of process models. Mining such hierarchical process models from event logs is an active subfield in process mining research. Existing approaches often require activities to be labeled with a hierarchy notion or context data to identify hierarchies. In this work, we propose the ProcessFragmentMiner (PFM) to fragment an event log into subprocesses that represent the root process. PFM harnesses the dependency matrix generated by the heuristics miner in combination with different fragment ranking mechanisms. This work uses the inductive and split miner to mine the resulting process models for each fragment. PFM is evaluated against a supervised and an unsupervised hierarchical process mining approach from the literature. We find that PFM works best in combination with the presented Bigram ranking method and can match supervised approaches for some data sets. The proposed PFM approach enables hierarchical process mining on any event log without the need for any preprocessing steps.