Publikationen

Autor	Reynolds, Steven Lamarr; Stromberg, Jonas; Lücke-Tieke, Hendrik; May, Thorsten; Kohlhammer, Jörn
Datum	2025
Art	Conference Paper
Abstrakt	Data science is integral. Its importance continues to grow, and so does the need for adequate tools to integrate multiple datasets and audit their transformations. In rapidly evolving fields where data formats frequently change, this task is often performed manually. However, manual data-wrangling tasks are often error-prone and time-consuming for human analysts. In this paper, we propose a semi-automatic data wrangling approach that allows analysts to interactively integrate heterogeneous structured datasets into a unified target data format. This is achieved by abstracting raw data into schemas and transforming relevant attributes into a target data schema. To assist analysts, it also suggests an initial transformation and visualizes the resulting data to verify the transformation. We also provide a use case to demonstrate the capabilities of our interface for wrangling and verification. Supplemental materials are available at https://osf.io/dscfb/?viewonly=7b50be799c8540eaaf50e5b296629530.
Konferenz	International Conference Information Visualisation 2025
Url	https://publica.fraunhofer.de/handle/publica/499187

Interactive Integration of Heterogeneous Datasets for Analytical Tasks