Applying Process Mining Algorithms in the Context of Data Collection Scenarios

Despite the technological progress, paper-based questionnaires are still widely used to collect data in many application domains like education, healthcare or psychology. To facilitate the enormous amount of work involved in collecting, evaluating and analyzing this data, a system enabling process-driven data collection was developed. Based on generic tools, a process-driven approach for creating, processing and analyzing questionnaires was realized, in which a questionnaire is defined in terms of a process model. Due to this characteristic, process mining algorithms may be applied to event logs created during the execution of questionnaires. Moreover, new data that might not have been used in the context of questionnaires before may be collected and analyzed to provide new insights in regard to both the participant and the questionnaire.

This thesis shows that process mining algorithms may be applied successfully to process-oriented questionnaires. Algorithms from the three process mining forms of process discovery, conformance checking and enhancement are applied and used for various analysis. The analysis of certain properties of discovered process models leads to new ways of generating information from questionnaires. Different techniques for conformance checking and their applicability in the context of questionnaires are evaluated. Furthermore, new data that cannot be collected from paper-based questionnaires is used to enhance questionnaires to reveal new and meaningful relationships.