Semi-automated Ontology Creation for Semantic Search in Business Process Exploration

In a globalized economy with lean production structures well documented processes are more important for companies than ever. But not all processes can be documented in advance. In complex processes, unforeseen errors can arise and have often to be dealt with in an improvised manner. And if these errors happen regularly, they might point out problems in the company's core processes. In order to find these problems one first has to analyze the aforementioned improvised error handling processes. If they occur regularly, it is likely that they hint at a problem. Finding these undocumented error handling processes is, however, not an easy task as they are undocumented. One strategy is to start with a single instance of such process and search for similar instances to generalize a process structure from these instances. On major problem is that most steps of these processes are manual and thus involve natural language in a context specific to each step of the process. This paper shows how ad-hoc ontology snippets can be generated from corporate data such as organization structures to tune automatic document processing for the analysis of each process step separately.