VIStory: interactive storyboard for exploring visual information in scientific publications
Wei Zeng,Ao Dong,Xi Chen,Zhang-lin Cheng
DOI: https://doi.org/10.1007/s12650-020-00688-1
IF: 1.7
2020-08-16
Journal of Visualization
Abstract:Many visual analytics have been developed for examining scientific publications comprising wealthy data such as authors and citations. The studies provide unprecedented insights on a variety of applications, e.g., literature review and collaboration analysis. However, visual information (e.g., figures) that is widely employed for storytelling and methods description are often neglected. We present <i>VIStory</i>, an interactive storyboard for exploring visual information in scientific publications. We harvest a new dataset of a large corpora of figures, using an automatic figure extraction method. Each figure contains various attributes such as dominant color and width/height ratio, together with faceted metadata of the publication including venues, authors, and keywords. To depict these information, we develop an intuitive interface consisting of three components: (1) Faceted View enables efficient query by publication metadata, benefiting from a nested table structure, (2) Storyboard View arranges paper rings—a well-designed glyph for depicting figure attributes, in a themeriver layout to reveal temporal trends, and (3) Endgame View presents a highlighted figure together with the publication metadata. We illustrate the applicability of <i>VIStory</i> with case studies on two datasets, i.e., 10-year IEEE VIS publications, and publications by a research team at CVPR, ICCV, and ECCV conferences. Quantitative and qualitative results from a formal user study demonstrate the efficiency of <i>VIStory</i> in exploring visual information in scientific publications.
computer science, interdisciplinary applications,imaging science & photographic technology