Human-Centered Data Science

In recent years, machine learning has developed rapidly, opening up new opportunities in a variety of societal, scientific and technological areas. However, experiences over the past few years have increasingly shown that the focus on purely statistical and numerical aspects in data analysis fails to capture social nuances, to take ethical criteria into account, or to consider the human in the decision making process. The research area Human-Centered Data Science closes this gap at the intersection between human-machine interaction (HCI), computer-supported cooperative work (CSCW), computation and statistics of data science. In this talk, I use recent examples from my research area on Human-Machine Collaboration to show how purely quantitative research results are not sufficient when considering socio-technical systems. In my discussion, I focus in particular on the principles of "Trace Ethnography", which are based on a researcher learning to interpret log data as part of the lived and learned experience of a community. As a result, in addition to the findings of the case study, I will outline what next steps (in research and training) need to be taken to enable Human-Centered Data Science.