At the end of October, thousands of people from around the world tuned in to watch the IEEE VIS Conference, the most significant annual conference on research and innovation in the field of data visualization. For the first time, the conference took place virtually and was broadcasted live on YouTube. Over 5,700 people registered for the conference, which included over 4,300 first-time participants, and many of these self-identified as being in countries that had never before been represented at VIS. The previous record attendance for an (in-person) VIS conference was 1,250 people, and at that time, most attendees were from North America, Western Europe, and East Asia. Despite the increase in participation, you may have noticed that the #ieeevis hashtag on Twitter was quieter this year; this is because the conference adopted a Discord instance for registered attendees, with associated channels for every YouTube stream.
There is an incredible amount of work presented at VIS each year, with up to eight hours of research paper presentations, panel discussions, tutorials, and workshops taking place across up to eight parallel tracks for six consecutive days. Needless to say, we only saw and participated in a small fraction of the conference, so this set of highlights are limited to this subset.
In this article, we offer a few highlights from several categories of content from this year’s conference, with links to watch the recordings of each highlight. Matt comments on this year’s award recipients, short and long research talks, the VIS Arts Program, and VisInPractice, while Noëlle comments on VizSec, VisComm, and Vis For Social Good.
Matt Brehmer’s VIS 2020 Highlights
This was my ninth VIS conference and my third as a member of the conference organizing committee. This perspective allowed me to reflect on what I found to be relevant, novel, and potentially impactful with respect to visualization research and practice in the years to come.
Recognition & Awards
Part of the conference’s opening session on Tuesday was designated for the celebration of and reflection on time-tested VIS research papers, those that have had lasting impact on both research and practice. 2020 marked the 20-year anniversary of the original Polaris paper by Chris Stolte, Diane Tang and Pat Hanrahan, and Chris reflected on the paper and its legacy: Polaris would become Tableau (5 min video).
A 10-year test-of-time award went to Edward Segel and Jeffrey Heer for their 2010 paper describing a design space for narrative visualization, which introduced the community to seven genres of narrative visualization. Both authors reflected on how it inspired researchers and practitioners to tell stories with data (5 min video).
Jumping ahead to the present, there were several awards presented to the authors of highly-reviewed papers submitted to this year’s conference, and of these I was particularly fascinated by the Anatomical Edutainer presented by Marwin Schindler; this project combines foldable papercraft visualizations with colored transparency filters or colored light, resulting in low-cost, yet interactive, representations of 3D volumes, in which filters or lights illuminate different substructures within the volume (10 min video).
Long Paper Presentations
Presentations of long (10-page) research papers are the primary event of the conference, and this year there were over 150 such presentations. These papers were either submitted to the conference review process in the spring of 2020, or they appeared in the IEEE Transactions on Visualization and Computer Graphics journal at some point over the last year. Most of the former will appear in a special issue of TVCG in January 2021, and preprints for many of these are already on publicly accessible preprint servers such as arxiv or OSF. Of the many paper presentations that I saw, I’ll point out a few that I found to be particularly interesting and inspiring.
Having contributed to the design of and reflection on expressive visualization authoring tools in recent years, I was happy to see that the boundary for such tools continues to be pushed by Theophanis Tsandilas with his StructGraphics system. Presented in a session on creating and sharing visualizations, StructGraphics delays the assignment of data bindings to vector graphics until late in the authoring process, and these assignments can come from multiple data tables and at varying levels of scope (12 min video).
A fascinating presentation in a session dedicated to libraries, toolkits, and systems pertained to arranging and interacting with small multiples, in which Fritz Lekschas presented a framework and library for “piling” images or visualizations in various small multiples configurations. (If you’re like me, you might be reminded of the BumpTop user interface from the mid 2000s). The piling paper was the recipient of an honorable mention for best long paper (12 min video | piling.js github).
Later in the week, a session on guidelines and design spaces provided back-to-back presentations about configuring multiple-view visualizations and comparative layouts, presented by Xi Chen and Sehi L’Yi, respectively. Since many visualization applications and dashboards incorporate multiple views on the data and are intended to support comparison tasks, both of these papers could be helpful resources for designers (2 x 12 min video).
Finally, I contributed to one long paper appearing at this year’s conference, which described QualDash, a dynamic dashboard specification application for healthcare quality audits in NHS hospitals; this paper was presented by my collaborator Mai Elshehaly in a session dedicated to the topics of health and disease (12 min video).
Short Paper (Four-Page) Presentations
The first short paper highlight is Encodable, a type-sensitive visualization grammar for building reusable visualization components by Krist Wongsuphasawat, presented in a session dedicated to systems, libraries, and algorithms. Encodable lets developers specify the data parsing and rendering logic and is compatible with various rendering systems. Krist’s paper received an honorable mention for best short paper (7 min video).
In 2013, Samuel Huron and colleagues introduced the concept of visual sedimentation, the animated accumulation of particles into aggregate marks, which is particularly useful when visualizing dynamic streaming data. At this year’s conference in a session on interaction and animation, João Moreira expanded upon this concept to evaluate a design space of animated transitions between streaming particles and various aggregate visualization techniques (e.g., heat maps, line charts, bar charts), culminating in a set of recommendations for designers of animated visualization (7 min video).
The last short paper that I want to mention is one with considerable relevance for 2020 and specifically for those working from home, collaborating with and presenting to remote colleagues and clients. The MeetCues application presented by Marios Constantinides is a video conferencing plugin that provides visual indication of engagement and attention among meeting attendees with an augmented back-channel. Presented in a session on text and communication, MeetCues was also one of the few mobile applications presented at the conference (7 min video).
VIS Arts Program
One of the events that I look forward to most at VIS each year is its Arts Program, which would typically include a gallery exhibition. Despite a virtual-only program, this year’s arts program did not disappoint: I’m consistently inspired by the work that I see exhibited and presented in this program, and I would urge researchers and developers to draw from the ideas and motifs explored by the artists and designers participating in this event.
Of particular note was the Tsuga Convictio project presented by Cathryn Ploehn, which incorporates deliberately ambiguous visual representations based on dendrological and mycological metaphors, with input data from a hand gesture recognition system. Specifically, the visual inspiration for this project was the old growth hemlock trees and their root networks of mycelium in the Western Allegheny Plateau. The components of Tsuga Convictio serve to reflect and complement conversations that co-located groups engage in, encompassing not only their words but their body language as well (15 min video).
Another notable arts program entry was the HeartBees project presented by Chao Ying Qin, which captures biometric data from groups of people (crowds, employees, audiences, etc.) and visualizes the inferred affect of crowds as animated biomimetic flocking boids (15 min video).
The arts program’s virtual exhibition included the Cangjie project by Weidi Zhang and Donghao Ren, a virtual reality experience that generates novel Chinese characters based on the VR headset’s camera orientation, drawing these in an immersive environment (10 min video).
Lastly, I’ll speak about VisInPractice, a VIS event that I’ve been co-organizing since 2018. This year I was joined by organizers Sean McKenna, Matt Larsen, and Zhicheng (Leo) Liu, and together we invited nine visualization practitioners to speak and/or appear as panelists, with six hailing from the Salt Lake City area (the VIS conference was originally planned to take place in SLC).
First, Alan Wilson spoke about visualization for design systems, drawing from his survey of style guides and from his experience at Adobe. I particularly liked his characterization of how insights from the visualization research community eventually find their way into design systems, albeit through an indirect and convoluted process (20 min video).
Next, Julia Silge of RStudio contributed an overview of visualization for machine learning, differentiating techniques that are useful across alternating phases model fitting, model tuning, and evaluation (20 min video). [Coincidentally, Julia’s talk took place within minutes of a very complementary presentation taking place in the Visualization in Data Science (VDS) session, in which my Tableau Research colleague Anamaria Crisan spoke about the current application of visualization and the opportunities for visualization innovation across data science and machine learning workflows.]
Finally, I had the privilege of hosting a lively and interesting panel discussion on what it’s like to be a visualization freelancer or consultant, in which I chatted and relayed audience questions to panelists Jane Zhang, Kristin Henry, and Curran Kelleher. The catalyst for this panel session was Jane’s Nightingale article: “How Self-Employed Data Visualization Designers Make a Living”. After a presentation by Jane based on that article, I chatted with the panelists about finding and retaining clients, negotiating project timelines and fees, and the concept of an overflow network, in which your fellow freelancers should not be seen as your competitors, but rather as people you can hire or parcel work out to when a project is too much for you to take on alone (90 min panel).
Matt Brehmer is a senior research staff member of Tableau Research in Seattle, where he focuses on information visualization and human-computer interaction. Learn more about his work at mattbrehmer.ca or follow him @mattbrehmer.
Noëlle Rakotondravony’s VIS 2020 Highlights
This was my third VIS conference and my first as a Student Volunteer.
In this reflection, I highlight some of the talks and discussions that will definitely impact my approach to research (especially to evaluation) as a PhD student doing data visualization.
VizSec is the conference’s Symposium on Visualization for Cyber Security. In his keynote, Joshua Saxe talked about one of the big challenges in this very practical field: “how to convince organizations to use VizSec research?”. In the practitioner’ talks that followed the keynote (video), Leo Meyerovich (Graphistry), Robert Gove (Two Six Labs), and Younghoo Lee (Sophos) discussed the different opportunities to achieve a bigger impact in VizSec research.
Another topic that was highlighted in the FAQs and discussions was the need to go beyond traditional user studies and perform “organizational studies” that measure the actual costs and benefits for organizations in adopting VizSec research outcomes. All of the ideas that emerged will certainly influence the design and evaluation approaches in future VizSec research.
VisComm (the Workshop on Visualization for Communication) took place on Sunday, Oct. 25th. The topics were interesting and covered studies of visualizations used in scientific communication, in medical communication with patients, and for communicating everyday life information like nutrition facts labels (10 min video).
One particularly interesting talk at VisComm was by François Lévesque about the evolution of taxes, benefits, and public spending for taxpayers in Canada. I enjoyed this workshop because all the presentations showed how visualizations can help disseminate research results and impact citizens’ lives (10 min video).
Visualization for Social Good
“How citizens benefit from data visualization research and practices” was at the core of the discussion during the Visualization for Social Good panel on Friday Oct. 29th. The panelists Catherine D’Ignazio, Ronald Metoyer, Michelle Borkin, and Evan Peck presented their work in design and data visualization and how these efforts are impacting local communities. Data visualization should serve as a tool for engaging audiences with data that reflect their own realities, or as a medium through which communities can own the outcomes of their projects through partnerships. There was also a discussion with attendees about how researchers can find a balance between publishable works and works that actually impact the community. Another topic covered was how to measure the value of social good in our research, or whether the impact on social good should be a part of the evaluation criteria of published works (90 min panel video).
Noëlle Rakotondravony is a PhD student in Computer Science. She focuses on cybersecurity data visualization and datavis for communication. Noëlle is also an editor at Nightingale. (LinkedIn | Twitter)
Next year, the IEEE VIS 2021 conference is slated to take place between October 23rd and 28th. Should a safe in-person event be possible by then, the conference will be in New Orleans. However, it’s more likely that it will be a virtual or hybrid event. While it’s sad to go so long without reconnecting with our friends and collaborators from around the world, the virtual conference has had a few positive repercussions: increased participation from first-time attendees, the ability to revisit and watch conference content asynchronously, and a more coherent conference backchannel experience.
In the meantime, there is plenty of VIS 2020 conference content to catch up on. If you participated this year, you probably watched an entirely different set of sessions than we did. Let us know what you enjoyed, or better yet, we encourage you to write about your own highlights as we have done here.