The process mining field supports the discovery of process models using audit trails logged by information systems. Several mining techniques are able to deal with unstructured processes, mainly through cluster analysis. However, they assume the previous extraction of an event log containing related instances. This task is not trivial when the source system doesn’t provide a reliable separation of its processes and allows the input of data through free text fields. The identification of related instances should, in this case, be explorative and integrated into the process mining tool used in later stages of the analyst’s workflow. To this goal, the MANA approach was developed, allowing the explorative selection and grouping of instances through a canonical database.