1 of 45

Accessible Flowcharts: A Feasibility Study with BVI Participants

Brianna L. Wimer

University of Notre Dame

Jennifer Mankoff

University of Washington

2 of 45

What to Expect in Today’s Session

Quick introductions of panelists
Each panelist will share their work, highlighting different approaches to making diagrams and visualizations accessible
Moderated discussion, exploring challenges, solutions, and diverse perspectives in accessibility
Audience Q&A, opening the floor for questions

3 of 45

The Accessibility Gap in Diagrams and Visualizations

Many digital and print visualizations and diagrams are not accessible to people with disabilities
Current accessibility solutions are often limited to alt text or are nonexistent, failing to provide meaningful access.
Different users require diverse access methods (text, tactile, auditory, interactive) to engage with visual information.

4 of 45

Meet the Panelists

Brianna L. Wimer

University of Notre Dame

Venkatesh Potluri

University of Michigan

Frank Elavsky

Carnegie Mellon University

5 of 45

Accessible Programming

Venkatesh Potluri�Assistant Professor

School of Information | University of Michigan

�

5

potluriv@umich.edu�https://blvi.dev

6 of 45

“Employment of data scientists is projected to grow 36 percent from 2023 to 2033, much faster than the average for all occupations.”

6

U.S. Bureau of Labor Statistics. Data Scientists: Occupational Outlook Handbook. Retrieved from: https://www.bls.gov/ooh/math/data-scientists.htm

7 of 45

Meet Dez

Blind software engineer.
Loves going on walks with friends, enjoys sighted guide sometimes.

7

8 of 45

Understanding Activity Recognition

8

What interesting parts in the dataset can you spot?

Did all persons wear the sensors similarly?

Bock et al . 2021. Tutorial on Deep Learning for Human Activity Recognition.

9 of 45

Alternative texts in a hundred K notebooks on the internet

9

10 of 45

Several programming domains and activities remain inaccessible.

10

11 of 45

Popular Libraries can’t embed ALT Text

11

12 of 45

MatplotAlt: A Python Library for Adding Alt Text to Matplotlib Figures in Computational Notebooks

Supports multiple strategies to generate and display alternative texts to Matplotlib figures with a single line of code
Generation

Generate_alt_text
Generate_api_alt_text

Display

In notebook
Command line

12

Download at https://blvi.dev/matplotalt

Kai Nylund, Jennifer Mankoff and Venkateshpotluri. Conditionally accepted, Eurovis 2025

13 of 45

MatplotAlt system

13

Matplotlib Figure

Figure features and data

generate_alt _text

Chart Type Classifier

Type

Image

generate_api_alt _text

Alt text

Data table

VLM alt text

add_alt_text

Methods:

Embed in html
Output as markdown or new cell
Export as image or text file

Heuristic alt text

14 of 45

PSST : Programable and interactive dashboard to explore streaming sensor data

14

Potluri et al. PSST: Enabling Blind or Visually Impaired Developers to Author Sonifications of Streaming Sensor Data. UIST’22. – blvi.dev/psst

15 of 45

Contributions

15

Accessible Programming Experiences

Accessible data representations

Data-driven accessibility studies

Impact

16 of 45

It is imperative that we ensure paradigm shifts in programming remain accessible!��Interested to collaborate? potluriv@umich.edu

16

17 of 45

Accessibility and Data Visualization

Frank Elavsky, fje@cmu.edu

hcii.cmu.edu, axle-lab.com, dig.cmu.edu

2025

frank.computer

Tool-making and guidelines

Hey everyone, I am Frank Elavsky, a PhD student and researcher at Carnegie Mellon University. I research, engineer, design, and contribute to standards at the intersection of accessibility and visualization.

So, I do work with practitioners a lot. And by practitioner here I mean someone who is not a student and not an academic researcher. These are people who do a specific kind of work in a specific area. For me, these are primarily data visualization and data science practitioners. And in my research, I think about people building and making visualizations all day long. The heart of what I do is about changing the behaviors and outcomes in the ways people build interactive data science systems, tools, applications, products, and services. I believe in this fundamental right that everyone should have: access to data. And also access to understanding the insights we produce from data. And people with disabilities have been excluded from this right. We communicate pandemics with charts and graphs, as well as the state of our elections. Personal informatics from health to finance rely on data visualizations. These artifacts are ubiquitous, virtually everywhere. I want these visualizations, and more importantly the information and decision-making that they enable, to be available for people with disabilities.

18 of 45

Industry and research-oriented visualization

18

Supporting data scientists

Visualizing a model of over 1m careers

My pre-phd work: supporting others with tools

19 of 45

Industry and research-oriented visualization

19

Supporting data scientists

Supporting scientists

Visualizing a model of over 1m careers

Shown here is Barry Barish giving the Nobel Lecture in

Physics in 2017 with my visualization

My pre-phd work: supporting others with tools

20 of 45

Industry and research-oriented visualization

20

Supporting data scientists

Supporting engineers

Supporting scientists

Visualizing a model of over 1m careers

Shown here is Barry Barish giving the Nobel Lecture in

Physics in 2017 with my visualization

Building a chart component library that supports 2000 devs and 200 products in 200 countries

My pre-phd work: supporting others with tools

21 of 45

(My recent research)

21

To any visualization toolkit

Structure

Input

Rendering

Chartability:

What are accessibility barriers?�

Data Navigator:

How do we build accessible visualizations?

What and how of visualization accessibility

So I have 2 major published projects so far in my PhD, Chartability and Data Navigator. Chartability is about catching barriers. It’s essentially a list of 50 heuristics, with tools and methods, that practitioners can use to find accessibility barriers in their own work. I made Chartability because certain types of work, like data visualization, isn’t easily covered by WCAG and other accessibility standards. Chartability took all that it could from standards and synthesized those guidelines and tests for visualization and then we filled the rest of the gaps with research and community practices.

Data Navigator is my ideal tool for solving some of the hardest problems involved in making data visualizations and interactive data systems more accessible. You can have any graphic, rendered in any way, and use Data Navigator to design a rich experience for folks who use assistive technologies, like screen readers or keyboard-based tech too.

22 of 45

Chartability has helped me audit and train others

22

Perceivable:

6 – Low contrast

57 - Content is only visual

50 - Color alone is used

3 - Meaningful elements can be distinguished

Operable:

54 - Interaction modality only has one input type

58 - No interaction cues or instructions

5 - Low contrast on interactive elements

4 - Keyboard focus indicator missing

4 - Complex actions have no alternative

18 - Target pointer interaction is too small

Understandable:

4 - Interactive context is not clear

6 - Metrics or variables are undefined

Robust:

275 - Does not conform to standards

82 - Semantically invalid

12 - Fragile technology support

Compromising:

54 - Information can only be reached through single process

61 - Information cannot be navigated according to narrative or structure

Assistive:

101 - Navigation and interaction is tedious

Flexible:

2 - User style change not respected

121 - User text adjustments are not respected

1 - Scrolling experiences cannot be adjusted or opted out of

Contrast and textures cannot be adjusted

978 access failures found in ~60 minutes.

23 of 45

Data Navigator opened up collaborations with already-experts

23

Interactive demo link

24 of 45

A hard problem: Access Friction is when accessibility for someone produces a barrier for others

24

25 of 45

What about this is accessible? Why?

25

26 of 45

What about this might be a barrier? Why?

26

27 of 45

Softerware: malleable systems with guardrails

27

Interactive demo link

28 of 45

Accessibility and Data Visualization

2025

frank.computer

Slides here

Frank Elavsky, fje@cmu.edu

hcii.cmu.edu, axle-lab.com, dig.cmu.edu

Tool-making and guidelines

29 of 45

ACCESSIBLE FLOWCHARTS: A FEASIBILITY STUDY WITH BVI PARTICIPANTS

Brianna Wimer

Kelly Avery Mack

Ronald Metoyer

Jennifer Mankoff

30 of 45

Engineering Design Process Flowchart

Flowcharts are important diagrams in STEM education…

Define Problem

Conduct Research

Brainstorm Solutions

Create a prototype

Test and evaluate prototype

Improve the design

but accessibility solutions are limited, often relying only on alternative text

31 of 45

OUR GOAL: INVESTIGATE THE FEASIBILITY OF THREE ACCESSIBLE FLOWCHART REPRESENTATIONS.

32 of 45

ACCESSIBLE FLOWCHART REPRESENTATIONS

Textual Summary With Q&A

33 of 45

ACCESSIBLE FLOWCHART REPRESENTATIONS

Textual Summary With Q&A

Interactive Navigable Representation

34 of 45

ACCESSIBLE FLOWCHART REPRESENTATIONS

Textual Summary With Q&A

Interactive Navigable Representation

Tactile Representation

35 of 45

User Study with 8 Blind and Visually Impaired (BVI) Participants

We evaluated comprehension accuracy and gathered usability feedback

Explored Each Representation

Answered comprehension questions

Shared Insights & User Experience

36 of 45

FINDINGS & ENHANCEMENTS: TEXTUAL SUMMARY + Q&A

Users struggled to form queries

Implement query suggestions for better interaction.

Well-structured but lacks spatial relationships

Introduce structured tables for nodes and edges

37 of 45

FINDINGS & ENHANCEMENTS: INTERACTIVE NAVIGABLE

Efficient for screen reader users

Challenges in understanding node relationships

Improve clarity for node relationships

Add notifications for loops and endpoints

38 of 45

FINDINGS & ENHANCEMENTS: TACTILE GRAPHIC

Optimize layouts for larger/more complex flowcharts

Slower navigation due to braille limitations

Explore touch-based interactions on tablets

Best for mental visualization

39 of 45

KEY DESIGN TAKEAWAYS

Multimodal approaches are essential

40 of 45

KEY DESIGN TAKEAWAYS

Multimodal approaches are essential

Flowchart Structure Knowledge Enhances Navigation

41 of 45

FUTURE WORK

Evaluate Hybrid Approaches

Develop Real-World Tools

42 of 45

FINAL THOUGHTS

Empowering users with multimodal access to flowcharts and other diagrams goes beyond basic textual descriptions. By expanding beyond static alternatives for various types of diagrams, we can create more inclusive tools that foster independence, comprehension, and deeper learning in STEM fields.

THANK YOU!

Brianna L. Wimer

bwimer@nd.edu

Kelly Avery Mack

kmack3@uw.edu

Ronald Metoyer

rmetoyer@nd.edu

Jennifer Mankoff

jmankoff@cs.washington.edu

43 of 45

Moderated Panel Discussion

44 of 45

Questions from the audience?

45 of 45

Thank You for joining today!

Brianna L. Wimer

bwimer@nd.edu

University of Notre Dame

Venkatesh Potluri

potluriv@umich.edu

University of Michigan

Frank Elavsky

fje@cmu.edu

Carnegie Mellon University

Jennifer Mankoff

jmankoff@cs.washington.edu

University of Washington