CGV: Small: Creating Information Visualizations without Programming

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

An ever-growing number of people and organizations find themselves awash in data. Virtually every domain has been affected by our increasingly powerful ability to monitor and record any kind of data that might be relevant. Unsurprisingly, a strong desire to better understand the data and communicate to others has accompanied this widespread access. Visualization is one technique that can be particularly effective for data analysis and presentation. The ability to represent data and its attributes in a visual manner often allows people to more easily grasp important characteristics of the data and to reach conclusions about its value. Similarly, people frequently prefer to represent data visually when explaining it to others. Unfortunately, to create a data visualization today, a person either needs to be a programmer and use libraries and toolkits developed for that purpose or the person needs to use (typically commercial) systems that provide limited visual representations. Neither solution is appropriate for data scientists who are not programmers but seek to design custom visualizations for the unique aspects of the data they explore. This research is developing new methods of creating visualizations without requiring the use of programming. An initial step of the research is identifying the core primitives of the data visualization process, both with respect to representation and interaction. The research is also formulating a theory and model of how to represent these primitives so that people can easily specify their design intent, thus providing an expressive framework for constructing visualizations. Ultimately, the framework is embodied through the development of a prototype visualization construction system that empowers users to create custom visualizations. Evaluations of the effectiveness of the primitives, the conceptual framework, and system complete the project. The results of this project are expected to have broad impacts in a variety of domains and applications. When people from a variety of fields such as journalism, business, science, and health are able to rapidly develop visualizations of their data, the disciplines themselves will benefit from stronger abilities to analyze data and communicate knowledge from the data. The ability to construct visualizations without programming also enables younger students who have not yet studied computer science to engage in data analysis and visualization design. This research will create learning modules about visualization and data analysis for local high schools and students with a goal to grow interest within the students for areas such as data science, visualization, and human-computer interaction. Additionally, this project provides education and research experience for graduate students. Recent noteworthy national research agendas have identified the need for more students and young researchers trained in STEM-related disciplines. The resulting visualization construction system being created will be freely available on the internet for anyone to use. All results of the research will be communicated through academic venues and the publicly-accessible website (http://www.cc.gatech.edu/gvu/ii/designvis).