EXAMPLE AIDED DESIGN: A PATH TO AUTOMATING EXPRESSIVE VISUALIZATION DESIGN
Hannah Bako
Monday, May 6, 2024, 2:00-3:00 pm 
Abstract
Abstract:
Data visualizations are a powerful tool for understanding data and conveying complex information to audiences with diverse technical backgrounds. However, it is not always obvious to designers what visual representation is appropriate for their design or analytical intent. When designers lack clear guidance for selecting appropriate visualizations, poor design choices may lead to the misrepresentation and misinterpretation of data. Such outcomes can have detrimental consequences for both businesses and society as a whole. When designers cannot develop design ideas, they often use examples of past visualizations to find inspiration or conceptualize the space of possible designs. Visualization examples offer an avenue for designers to understand what designs are possible for the data they are working with and how they can author these designs. Yet, visualization research has completely overlooked the use of examples for visualization design exploration and generation.
This proposal focuses on the use of examples to support the exploration of relevant visual representations as well as the automatic generation of selected designs. The first phase of the proposal presents work that establishes a holistic understanding of the practices and processes involved with the search, identification, and utilization of examples through an interview study. We follow this thread of research with a controlled experiment to quantitatively measure the factors that modulate what types of examples designers select and the ideas they incorporate into their visualization designs. The second part of this proposal seeks to develop computational techniques that harness valuable design knowledge within examples to build systems that support the exploration, identification, and refinement of potential designs. We present a computational approach that leverages D3 code templates to support code recommendation and augmentation, allowing the rapid prototyping of complex interactions in visualization designs. Finally, we propose a novel approach built on our conceptual framework of example usage to support the automatic generation of diverse data visualization designs in response to natural language queries made by designers.
This proposal focuses on the use of examples to support the exploration of relevant visual representations as well as the automatic generation of selected designs. The first phase of the proposal presents work that establishes a holistic understanding of the practices and processes involved with the search, identification, and utilization of examples through an interview study. We follow this thread of research with a controlled experiment to quantitatively measure the factors that modulate what types of examples designers select and the ideas they incorporate into their visualization designs. The second part of this proposal seeks to develop computational techniques that harness valuable design knowledge within examples to build systems that support the exploration, identification, and refinement of potential designs. We present a computational approach that leverages D3 code templates to support code recommendation and augmentation, allowing the rapid prototyping of complex interactions in visualization designs. Finally, we propose a novel approach built on our conceptual framework of example usage to support the automatic generation of diverse data visualization designs in response to natural language queries made by designers.
Examining Committee
Bio
Hannah Bako is a PhD student at the university of Maryland, College Park in Computer Science advised by Dr. Zhicheng Liu. Hannah works at the intersection of Data Visualization and Human Computer Interaction investigating novel techniques to support the visualization design process. She holds an MSc in Software Engineering from Stevens Institute of Technology and a BSc in Computer Information Systems from Babcock University.
This talk is organized by Migo Gui