At the Human-Centered Computing (HCC) research group, I was employed in the publically funded IKON project, collaborating with a major German natural history institution. The goal of the project was to develop a data visualization for non-technical experts (see final prototype on the left), which was meant to (1) help researchers and staff working at the institution understand ongoing research and knowledge transfer, and (2) promote collaboration and knowledge exchange beyond hierarchical ‘silos’.
My role was not only to head the project regarding communication, financial controlling, knowledge dissemination and research direction, I was also the primary researcher and the only one applying qualitative methods. Accordingly, I conducted contextual inquiries, explicitation interviews, and prototype testing. From this qualitative research, I derived that an ML-driven visualization was best suited for the use case by showing thematic similarities. This led me to study the research fields of interpretability and explainable AI, and conclude that a participatory design workshop with potential explanations for our prototype was required. Accordingly, we deployed a co-design method (see below or here), to find out how exactly our stakeholders reasoned about the proposed visualization, a paper on which has recently been accepted for ACM GROUP 2022.
Find details on the development of the ML-driven visualization and co-design workshop below:
- Sense-Making of Machine Learning by Non-ML expertsSense-Making of Machine Learning by Non-ML experts Part of Project IKON To understand how stakeholders would make sense of the proposed prototype, we developed a method for conducting co-design workshops […]
- Visualizing Machine Learning UncertaintyVisualizing Machine Learning Uncertainty Part of Project IKON From qualitative research I conducted, we concluded that our use case called for an ML-driven visualization, as this could show stakeholders potential […]