Irrelevant Sentences Detection for Automated Business Process Modeling

In Business Process Management, a full automation of the modeling process can reduce the resource requirements by up to 60%. Current work propose solutions only working when process descriptions are sequential and do not contain noise like irrelevant information. Because process descriptions are unstructured, described on a meta level, it may contain both relevant and irrelevant information; parts of the textual descriptions might be irrelevant for the generated process model, and it is important to identify and ignore them. Interestingly, no study addressing this problem was found, although this provide the starting point for the construction of a modeling engine.