Disclosed are systems, methods, and machine-readable storage media for converting process data extracted from one or more data source systems into a data-source-independent intermediate representation, and then applying a domain-specific semantic ontology to the intermediate representation to create a semantic representation of the process data. The intermediate representation may specify, for each instances of a process object within a process flow, a unique identifier, a set of observations, and references to process-object instance immediately preceding or following the process-object instance at issue.