Explore visualizations by memorability score
×
![]()
Scientific Visualization Memorability Dataset
We report results from a preliminary study exploring the memorability of spatial scientific visualizations, the goal which is to understand the intrinsic visualization features that contribute to memorability. The evaluation metrics include three objective measures (entropy, feature congestion, the number of edges), four subjective ratings (clutter, the number of distinct colors, familiarity, and realism), and two sentiment ratings (interestingness and happiness), all inspired by recent memorability studies in vision science and infographics. By constructing a scientific visualization dataset and conducting an online experiment, we collect memorability scores on Amazon Mechanical Turk (AMT) of 1142 images from the original 2231 images in published IEEE SciVis papers from 2008 to 2017. Results showed that the memorability of scientific visualizations is correlated with color and layout. We further investigate the differences between scientific visualization and infographics as a means to understand memorability differences of data attributes.
Explore visualizations by memorability score
Memorability is often studied by presenting a brief glimpse of images, and features at all levels, low to high, can contribute to their comprehension. Low-level features are used to describe image elements such as color statistics and luminance. High-level features containing Gestalt groupings can help humans interpret objects in complex spatial data visualization. Since both low-level and high-level features are used extensively in spatial visualization, understanding the correlation of image metrics with these features could automate the design of visualization techniques and increase the impact of visualization in real-world applications. In our study, we adapt and expand recent vision science studies and visualization methods and measured three objective (entropy, feature congestion, and the number of edges), four subjective (clutter, number of distinctive colors, familiarity, and realism), and two affectiveness metrics (interestingness or happiness) metrics to investigate the relationship between memorability and visualization features.
The following tools provide a convinient approach to observe the correlation betweens above features. Each dot in the left plot represents one visualization. Hovering over each dot shows visualization associated with each point. We also provide a comparison option with the Massive dataset.
We shared the following datasets and code of our experiment. Please cite our paper if you find these datasets useful in your research.
Recent Data Change:
Acknowledgement
We thank Drs. Jeremy Wolfe and Aude Oliva for discussions on memorability and the anonymous reviewers for their constructive comments. This work was supported in part by NSF IIS-1302755, CNS-1531491, and DBI-1260795 and by NIST MSE-70NANB13H181. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of National Institute of Stan- dards and Technology (NIST) or the National Science Foundation (NSF.)
