A Complex Event Recognition (CER) system takes as input a stream of events, along with a set of patterns, defining relations among the input events, and detects instances of pattern satisfaction, thus producing an output stream of complex events. Typically, a CER system needs to handle high-volume and high-velocity data streams and must be able to do so with strict latency requirements. It is therefore crucial for such systems to be extremely efficient. Besides the issue of efficiency, there are two other complementary issues: that of extracting from an input stream a set of interesting patterns via Machine Learning techniques; and that of predicting the occurrence of complex events even before they happen. Process Mining is a field related to CER that analyzes business processes and attempts to extract a high-level description of them, typically in the form of a transition system. Despite the fact that CER and Process Mining share many common characteristics, their relation has not been yet investigated. The goal of this thesis is to examine which Process Mining techniques (and to what extent), both for learning and prediction, may be useful for CER.