Abstract
Inspecting the high-dimensional data underlying projection spaces requires an appropriate exploratory visual analytics approach. However, prior work often relies either on tedious manual selection of items and groups or on showing all details at once, which scales poorly and overwhelms users. We present an adaptive and adaptable visual analytics tool for sensemaking in complex high-dimensional spaces. Our tool provides a flexible set of parameters to dynamically determine what to show, where to show it, and at what level of detail. By combining user interaction with automatic interest computation and clustering, it produces continuously updated, map-like visualizations in which relevant structure is emphasized and contextual detail appears on demand across diverse datasets and domain-specific detail views. The interface is adaptive in how it updates annotation density, aggregation, and detail level, and adaptable in how it exposes user selections, queries, navigation, and parameter controls for steering this behavior.
Citation
Christian A.
Steinparz,
Yannick Metz,
Andreas
Hinterreiter,
Manuela Waldner,
Mennatallah El-Assady,
Marc
Streit
Parametric Detail-on-Demand for Projection Spaces of High-Dimensional Data
Joint Proceedings of the ACM UMAP Workshops 2026, UMAP 2026, June 8–11, 2026, Gothenburg, Sweden,
4206:
249-252, 2026.
BibTeX
@inproceedings{2026_umap_paraadapt_projection_spaces,
title = {Parametric Detail-on-Demand for Projection Spaces of High-Dimensional Data},
author = {Christian A. Steinparz and Yannick Metz and Andreas Hinterreiter and Manuela Waldner and Mennatallah El-Assady and Marc Streit},
booktitle = {Joint Proceedings of the ACM UMAP Workshops 2026, UMAP 2026, June 8–11, 2026, Gothenburg, Sweden},
publisher = {CEUR-WS.org},
address = {Gothenburg, Sweden},
url = {https://ceur-ws.org/Vol-4206/ParaAdapt-2.pdf},
volume = {4206},
pages = {249-252},
month = {May},
year = {2026}
}
