Due to increased computerization and automation of business processes, the amount of available process data increases, which can make it more difficult to maintain a real-time overview of running processes and detect trends and developments in lager volumes of historical process execution data. Data visualization techniques are well-established in real-time monitoring and retrospective analysis of business process data, and these visualization techniques are typically combined with automated data processing, machine learning, and statistics. However, visualization poses certain limitations and challenges, such as the difficulty to focus on other work while performing visual process monitoring, or the limited number of visual dimensions onto which data can be mapped. To alleviate some of these challenges, we propose to combine visualization and visual analytics with sonification (the representation of data using non-speech sound) to multi-modal systems. Sonification has several characteristics that make it suitable for monitoring and analyzing business process data, such as the possibility to monitor passively or human ability to detect even slight deviations in rhythm and pitch over time. However, as its application to the domain of business processes has hardly been researched, many open research questions remain in this area. This PhD thesis answers several of those open questions, e.g. how to effectively combine visualization and sonification to multi-modal systems for business process monitoring and analysis, and which sonification techniques and mappings are best suited for the tasks and requirements at hand. To this end, the SoProMon system, a modular evaluation framework for multi-modal sonifications for process monitoring as a secondary task, has been developed. A quantitative and qualitative experiment based on this system has been devised. It aims to test the SoProMon framework and determine how different types of sonification can help in guiding attention. One of the main results is that a variant combining visualization, auditory alerts and continuous sonification increases monitoring performance significantly over visualization alone. Another developed prototype combines both modalities to support users in the post-hoc analysis of business process execution data. While it seems that a tight integration of both modalities can support users in post-hoc data analysis as well, this will still need to be proven through qualitative and quantitative experiments. Overall, this thesis proves that sonification can be a valuable supplement to visualization in many areas, especially those involving real-time monitoring.

Top