Authors: Dustin Arendt, Megan Pirrung
Abstract: Storylines are adept at communicating complex change by encoding time on the x-axis and using the proximity of lines in the y direction to represent interaction between entities. The original definition of a storyline visualization requires data defined in terms of explicit interaction groups. Relaxing this definition allows storyline visualization to be applied more generally, but this creates questions about how the y-coordinate should encode interactions when an this is tied to a particular place or state. To answer this question, we conducted a design study where we considered two layout algorithm design alternatives within a geo-temporal analysis tool written to solve part of the VAST Challenge 2014. We measured the performance of users at overview and detail oriented tasks between two storyline layout algorithms. To the best of our knowledge, this paper is the first work to question the design principles for storyline visualization, and what we found surprised us. For overview tasks with the alternative layout, which has a consistent encoding for the y-coordinate, users performed significantly better (p<.05) than the storyline layout based on existing design constraints and aesthetic criteria. Our empirical findings were also supported by first-hand accounts taken from interviews with multiple expert analysts, who suggested that the inconsistent meaning of the y-axis was misleading. These findings led us to design a new storyline layout algorithm that is a ``best of both'' where the y-axis has a consistent meaning but aesthetic criteria (e.g., line crossings) are considered.