Building novel interactive visualizations without programming

Team Members: Ramik Sadana, Chad Stolper, John Thompson, John Stasko

     Data Illustrator: Augmenting Vector Design Tools with Lazy Data Binding for Expressive Visualization Authoring - CHI 2018 (Best Paper)
    Atomic Operations for Specifying Graph Visualization Techniques - IEEE InfoVis 2017 Poster
    Expanding Selection for Information Visualization Systems on Tablet Devices - ACM ISS 2016
    Understanding Data-Driven Visual Encodings through Visual Deconstruction - IEEE InfoVis 2016 Poster
    Visualizing Social Media Content with SentenTree - TVCG (IEEE InfoVis 2016)
    Designing Multiple Coordinated Visualizations for Tablets - Computer Graphics Forum (EuroVis 2016)
    Emerging and Recurring Data-Driven Storytelling Techniques: Analysis of a Curated Collection of Recent Stories - Microsoft Research Tech Report
    Hands-On, Large Display Visual Data Exploration - IEEE InfoVIs 2015 Poster
    GLO-STIX: Graph-Level Operations for Specifying Techniques and Interactive eXploration - TVCG (IEEE InfoVis 2014)
    OnSet: A Visualization Technique for Large-scale Binary Data - TVCG (IEEE InfoVis 2014)
    Value-Driven Evaluation of Visualizations - ACM BELIV 2014

    VIDEO: Data Illustrator, CHI '18 (vimeo)
    VIDEO: Expanding Selection on Tablet Devices, ISS '16 (92 MB mp4)
    VIDEO: Tablet MCV vis (60 MB mp4)
    VIDEO: GLO-STIX, InfoVis '14 (30 MB mp4, YouTube link)
    VIDEO: OnSet (54 MB mp4)

Project Goals:
An ever-increasing number of people and organizations find themselves awash in data with a desire to better understand it and communicate it to others. Virtually every domain has been affected by our increasingly powerful ability to monitor and record any kind of data that might be relevant. Widespread computational resources, memory, and networking make the acquisition of data arguably easier than it ever has been.

Unsurprisingly, a strong desire to better understand the data and communicate to others has accompanied this widespread access. In domains such as business, health, science, and engineering, people seek easier ways to extract knowledge and insight from their data. They also need to present the data and their findings to others in a comprehensible and effective manner. Many such "data scientists" desire easier ways to gain a broad understanding of data from their world and to communicate it to others.

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 an information visualization today, a person either needs to be a programmer and use libraries and toolkits developed for that purpose or 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.

The goal of this project is to enable people who are not programmers to take advantage of data visualizations without requiring programming. One specific sub-goal is to develop a methodology and system that will support non-programmers to create sophisticated visualizations without any programming. A related sub-goal is to explore specific types of data and analysis contexts where we can provide visualization solutions while requiring little or no programming expertise.

Current results:
We have analyzed a wide collection of visualizations created "in the wild" to understand the mappings from data to visual elements employed in each. Our findings are driving the development of a framework for defining data-to-visual mappings, a type of visual grammar. We are developing a web-based tool to catalog these visualizations and the mappings they use. The framework is then, in part, serving as the foundation for the creation of a system to allow non-programmers to design and create visualizations of data sets they have. The system is called and we have made it available for all to try out and use. Our CHI 2018 paper explains the framework and describes how Data Illustrator works.

In a suite of sub-projects, we are exploring the creation of visualization tools, applications, or systems that provide data analysts with visualization capabilities and do not require programming, or they minimize the amount of programming required.

In one sub-project, we focused on set data. We developed the OnSet visualization technique that portrays large collections of binary set data. The technique supports exploration of large sets, in particular, operations like intersection and union across these sets. It uses a matrix-based visualization technique paired with interactive specification of those operations. To learn more about this approach, please see the OnSet project home page that includes an interactive implementation of the technique where you can upload your own data.

In a second sub-project, we are examining the simplified design and creation of graph and network visualizations. This has led to the GLO (Graph-Level Operations) framework in which we are identifying the primitives of network visualization. These primitives can then be made available to data scientists to promote simplified creation of network visualizations. For more on this idea, please see the GLO project home page.

In another sub-project, we are focusing on text and document data, more specifically, large collections of short social media postings. We have developed a visualization technique called SentenTree that provides an interactive visual overview of the key themes and opinions from a collection of social media messages. For more on this idea, please see the SentenTree home page.

We also have explored the development of visualizations for tablet computers, where one does not have the traditional keyboard and mouse input devices but instead must rely on touch input via a finger or stylus. We are designing and developing a multiple-coordinated view visualization system that can be totally controlled by multi-touch input. More about this sub-project, please see the Touch web page.

This material is based upon work supported by the National Science Foundation under Grant No. IIS-1320537. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Project information:
Award number: IIS-1320537
Title: CGV: Small: Creating Information Visualizations without Programming
Duration: 9/1/13 - 8/31/16 (extended to 8/31/17)
PI: John Stasko
Students: J. Alex Godwin, Ramik Sadana, Chad Stolper, John Thompson, Andrew Dai
Last updated: April 16, 2018