Data is ubiquitous in industrial, scientific and social processes. Data is collected as observation of natural phenomena, data is created from simulation models, data is used to guide strategic decisions, and data has a significant influence on the quality of our daily life. However, in all of these use cases, raw data is only necessary but not sufficient for human understanding of complex processes. We still lack advanced machine learning technology that does not end with the execution of algorithms but goes one step further and assists humans in data understanding. With our computer science research, we enable such data-driven approaches to assist decision makers by focusing on extensive exploitation of big data. We study new machine learning concepts that allow for scalable processing of heterogeneous data repositories, (semi-)automated knowledge discovery in complex data, and interactive data exploration. In all these research fields, we focus on the inclusion of domain expertise into our machine learning methods. This allows humans to steer data analysis to novel data-driven hypotheses and comprehensive understanding of their data.