Abstract
This January 2020 issue of the IEEE Transactions on Visualization and Computer Graphics (TVCG) contains the proceedings of IEEE VIS 2019, held in Vancouver, BC, Canada. Marking its 30th anniversary, IEEE VIS remains the premier venue for the visualization community, comprising three concurrent conferences: Visual Analytics Science and Technology (VAST), Information Visualization (InfoVis), and Scientific Visualization (SciVis). This preface outlines the organizational structure, the rigorous peer-review standards, and the key research trends—such as Explainable AI and natural language interfaces—that defined the 2019 program.
Introduction
IEEE VIS 2019 marked a significant milestone: three decades of advancing the science and practice of visualization. As the field has matured, it has increasingly intersected with other domains, most notably machine learning and human-computer interaction. This editorial overview synthesizes the contributions of the three core conferences, highlighting how the community is addressing the challenges of data complexity and the need for more transparent analytical systems.
The Three Pillars of IEEE VIS
The conference is structured around three distinct yet synergistic tracks, each catering to different aspects of the visualization landscape:
- VAST (Visual Analytics Science and Technology): Focuses on the integration of automated analysis with interactive visualizations to support complex analytical reasoning.
- InfoVis (Information Visualization): Concentrates on the design, theory, and evaluation of visual representations for abstract data.
- SciVis (Scientific Visualization): Addresses the visualization of physical and spatial data, often involving large-scale simulations in fields like medicine and physics.
A Rigorous Review Process
To maintain the high standards associated with TVCG, a two-cycle review process was employed. In the first cycle, each submission received four reviews—two from the International Program Committee (IPC) and two from external experts. Borderline cases were subjected to further deliberation by the Paper Chairs. Conditionally accepted papers then entered a second cycle, where revisions were verified by primary reviewers to ensure all technical and presentation concerns were resolved.
The 2019 selection process remained highly competitive, with acceptance rates hovering around 25% across all tracks:
- VAST: Received 170 submissions; 42 were accepted for the TVCG track (24.7%).
- InfoVis: Received 212 submissions; 53 were accepted (25%). Notably, the InfoVis program committee achieved 40% female representation, reflecting a strong commitment to diversity within the leadership.
- SciVis: Received 100 submissions; 25 were accepted (25%).
Emerging Themes and Key Insights
The 2019 proceedings reveal several dominant themes that reflect the current trajectory of the field:
The Rise of Explainable AI (XAI)
A pervasive theme across all tracks is the use of visualization to demystify complex machine learning models. Researchers are developing tools to make "black box" algorithms more transparent, fair, and interpretable, enabling users to understand the logic behind model predictions.
Natural Language and Intuitive Interfaces
The VAST track highlighted a shift toward more accessible interfaces. The award-winning FlowSense system, for instance, demonstrates how natural language processing can allow users to construct data workflows without needing deep technical expertise in UI widgets or coding.
Deep Learning in Scientific Workflows
In the SciVis track, deep learning is being leveraged for more than just classification. Systems like InSituNet use deep image synthesis to assist in the exploration of parameter spaces for massive ensemble simulations, facilitating data interpolation and view generation.
Focus on the Design Handoff
The InfoVis track emphasized the importance of the visualization process. Research such as "Data Changes Everything" explored the practical challenges of the handoff between designers and developers, acknowledging that the collaborative workflow is as vital as the final visual output.
Award-Winning Contributions
The 2019 committees recognized several papers for their exceptional originality and impact:
- VAST Best Paper: FlowSense: A Natural Language Interface for Visual Data Exploration within a Dataflow System by Yu and Silva.
- InfoVis Best Paper: Data Changes Everything: Challenges and Opportunities in Data Visualization Design Handoff by Walny et al.
- SciVis Best Paper: InSituNet: Deep Image Synthesis for Parameter Space Exploration of Ensemble Simulations by He et al.
Conclusion and Future Directions
As IEEE VIS enters its fourth decade, the community faces new challenges. The editors call for continued "deep reflection" on the theoretical foundations of visualization and a greater emphasis on interdisciplinary support. While the field has made significant strides in handling large-scale data, the focus is shifting toward supporting "sense-making" in the face of increasingly complex and opaque computational models. The 2019 proceedings serve as both a record of achievement and a roadmap for the future of visual analytics and information design.
Summary updated