Business Process Mining and Rules Detection for Unstructured Information

In this article we show how to find evidence of incomplete or fractured processes in non-structured reports of known business processes, by means of rules, patterns and detection of cause-effect relationships. A priori classifications and probabilities of process activities are used as inputs for the analysis and rules detection. In this method we use a domain-specific ontology associated to process activities in order to improve on previous results, where occurrence of a process in a document set was detected by means of SLM.