Making Computational Notebooks Accessible

In CS classes, we often use computational notebooks (e.g., Google Colab) to provide interactive learning experiences to implement and test algorithms. These notebooks are interactive, allowing students to run code snippets and see the results immediately. However, these notebooks are not accessible to screen reader users (at least for now). This section shows you how to make computational notebooks accessible.

To make Colab assignments more accessible, you should


Organize your computational notebook

Organizing your Colab activity with clear divisions is helpful for all students to make sense of the notebook and is especially helpful in preparing accessible content. This section provides an example of how to do that.

1. Check the table of contents.

On your Colab left bar, make sure each section has a clear title and description.

This is a screenshot of the Google Colab interface. On the left is a table of contents of the document (The table has two sections. The first section is "CSE 547 - Colab 0", which has a subsection "Setup." The second section is "Vietnam War" which includes four subsections of questions); On the right, it has notebook including Python code snippets and texts.

2. Divide the Colab document into separate Python scripts.

Migrate the sections into separate Python scripts. The goal here is that screen reader users can work on the modular script for each section.

This is a screenshot of a directory on github. It includes a folder (data), one markdown file (question.md), and five Python files (intro.py, section_1.py, section_2.py, section_3.py, section_4.py).

You might realize that we created a separate data folder in the GitHub directory compared to drive = GoogleDrive(gauth), which is unique to Google Colab. Make sure that you share your data file (e.g., Google Drive) so that everyone can have access to it.

3. Run the notebook to make sure everything works.

Always run the notebook before releasing the homework. Also, host your notebook in a shared environment that can be tested by the course staff ahead of time since library incompatibilities can present significant barriers to students.


Set up your server environment

Host your Colab notebook as Python scripts. The students can work on a server (e.g., CSE servers like attu1) or locally. We recommend a shared environment that can be tested by the course staff ahead of time since library incompatibilities can present significant barriers to students.

Working outside of interactive notebook environments means that students with visual impairment need to re-run entire programs rather than just single snippets/cells. This can take substantially more time, especially when the program is slow (e.g., ML on larger datasets).

This section shows you how to set up your server environment using the CSE server as an example.

1. Contact IT Support

Contact support@cs.washington.edu to create a folder in attu1 and add r/w permissions to the students based on their NetID. You can set up your homework/Colab code there. The students can also work on them there. Don’t use UW cycle servers; you will run into permission issues.

Sometimes, attu1 can be slow. Please contact IT support to set up a new server (e.g., AWS, Google Cloud) if you need a more stable server for class purposes.

2. Once you have the folder set up, git clone your homework/Colab code there.


Additional resources