Recommendations for the next process activities can solely be given for unstructured and partially for semi-structured processes. Unstructured and semi-structured processes can be characterised as Knowledge Intensive Processes (KIP). Such processes are based on subjective perceptions. The key driver of these processes is knowledge. When a process executor is unfamiliar with the process, recommendation support is possibly helpful dur-ing the process execution. For recommendations, which rely on objective facts, a process-aware recommender system (PARS) might assist. Such systems aim to predict how the execution of process instances is going to evolve. Based on these predictions, the most appropriate recommendations for the next activities can be provided. PARS is used in the field of process mining. Process mining is a research field positioned between machine learning and data mining, as well as process modelling and analysis. Process mining is distinguished between online and offline. Online process mining techniques focus on cases which are still ongoing. Here, the running cases can be influenced without changing the process. Recommendations are part of the online process mining activities. For fact-based recommendations, a model is learned from historical (post mortem) event data. In addition, streaming event data (pre mortem) is taken into account.