In information visualization research, little is known about how visualization itself can support the collection and modification of data. Despite numerous existing examples across everyday, public, and highly specialized contexts and a variety of domains (including but not limited to civic participation, decision making, data journalism, data analysis, forecasting, and health tracking), information visualization research has largely overlooked these types of visualizations, and instead predominantly focused on visualizing existing datasets. Most contemporary visualization models, tools, and guidelines assume that the data is known and available prior to creating a data visualization.

Input Visualization is a new emerging research area in information visualization that conceptualizes these visualizations as "visual representations that are designed to collect (and represent) new data rather than encode pre-existing datasets". By intertwining data collection and visualization, this approach challenges established assumptions in visualization research, where data is typically treated as pre-existing and separate from its visual representation.

This new lens of input visualization reorients visualization research and practice to consider a vast and underexplored design space in visualization, offering a novel approach to data collection. Considering visualizations as input devices opens new opportunities for data-informed decision-making, analytical workflows, and sensemaking both synchronously and asynchronously. Given the established power of visualization to communicate, explore, and analyze data – extending its role to include data collection further expands its potential. This is particularly relevant for applications such as computer-supported deliberation, social activism, data journalism, and data analytics. Input visualization is also a promising emerging technique for large-scale civic engagement, data analysis, and hacktivism, as it enables a large number of participants to collectively collect and represent data in real time. Moreover, it promises future opportunities for governments, NGOs, and private organizations to collect and analyze multidimensional opinion data in real time, supporting new forms of social reflection, social behaviors, and collective intelligence.

Despite its broad potential, input visualization remains a nascent area, requiring the development of new conceptual models, input-specific interaction techniques, tools, and guiding frameworks. Key challenges remain underexplored, including issues of scalability, socio-psychological biases in data, the design of input procedures, and ethics and ownership, as well as others to be identified during the seminar.

In this Dagstuhl Seminar, we aim to further develop and articulate the potential of input visualization by bringing together researchers from multiple disciplines, including information visualization, human-computer interaction, and political science, as well as practitioners working with interactive and data-driven systems. In this seminar, we will aim to explore the following topics:

• Input Mechanisms: Can we design and implement new visual data collection methods?
• Application Areas: How are input visualizations used at present? What is their future potential?
• Scalability and Bias: What are key challenges for collecting data at scale? What are methods for addressing bias?
• Representation, Exclusion, and Ethics: What are ethical challenges when using input visualizations in public arenas or civic engagement? How does this translate to design and implementation?

Through a combination of discussion and hands-on exploration with researchers and practitioners, the seminar aims to strengthen the input visualization community, refine current knowledge gaps, scope key research challenges, and set an agenda for future collaborative efforts.

Creative Commons BY 4.0

Derya Akbaba, Nathalie Bressa, Samuel Huron, Kim Sauvé, and Wesley J. Willett