Website Accessability by persons who are using screen readers

Website Accessability by persons who are using screen readers

User Story

As a blind person or other neurodiverse individual who has accessibility challenges with visual text or images, and uses a "screen reader" to interact with Dataiku Web Sites (Community, Learning, the Product itself).  I would like to be able to successfully use all web interfaces provided by Dataiku.  

COS

Note

  • In the United States, there is the Americans with Disabilities Act, that mandates that employers make accommodations for diverse workers.
  • Screen Reader turns visual images of a web site into spoken words.  
  • It is often important that graphics have alt tags.  For example in some places on the community site the Kudo thumb up and down options don't say that the thumbs-up icon is about setting a Kudo.
6 Comments
Mark_Treveil
Dataiker
Dataiker
Status changed to: Acknowledged

We acknowledge the need to meet accessibility standards in all our software. 

A plan to improve DSS accessibility is currently being discussed.

tgb417
Neuron
Neuron

Excellent.  Looking forward to hearing more about your progress.  

Mark_Treveil
Dataiker
Dataiker

I have quite of lot of experience in this field from a previous life, but equally it is always valuable to get direct feedback from disabled users.  Do you have any in your organisation, by chance, using DSS today?

tgb417
Neuron
Neuron

I agree that having a full time user of these tools is very helpful.  At this time I do not know a full time data scientist who is blind.  

As a dyslexic I do use screen readers from time to time. However, I only use a screen reader for taking in long passages of text for example in the academy, and some times in the community.  

This is how I noticed the problems with kudos on the community site.  

Then as a disabilities advocate, and a person with some knowledge of adaptive technology, I felt it important to raise the topic.  

If you need some assistance testing, I’d be willing to pitch in as I have time.

However, to your point recruiting a group of neuro and sensory diverse individuals as an advisory team would be very helpful to your efforts.  

That said I suspect that we may have a bit of a chicken and the egg problem here. Until DSS is well suited to screen readers it may be difficult to recruit a dedicated advisory team. 

In this thread I note that there is a blind R users group.

https://www.researchgate.net/post/Do_you_know_of_statistical_analysis_software_that_is_accessible_fo...

Here is the link to the blind r users group.

https://www.nfbnet.org/mailman/listinfo/blindrug_nfbnet.org

This looks to be out of the US National Federation of the Blind.  I would imaging other groups in the UK and Australia as well.  

I would think that when you are ready you may be able to recruit some more testers.  

tgb417
Neuron
Neuron

@Mark_Treveil 

A great example today of where Dataiku is missing meaningful alt tags is on the academy pages. For example.

https://academy.dataiku.com/path/ml-practitioner/intro-to-machine-learning-ml/634976

There are lots of images that have alt tags.  However, they add nothing to the understanding of the page.  They are set to exactly what the source is.  They do not describe what the image is exactly.  

A picture showing an inspector view of an Dataiku academy web page that has as an alt tag exactly the same as the source.  That is a relative path to the location of the image. tA picture showing an inspector view of an Dataiku academy web page that has as an alt tag exactly the same as the source. That is a relative path to the location of the image. t

 

 

Attached is a video in which you can hear what a user of the Apple Macintosh built-in screen reader sounds like when someone using a screen reader tries to listen to the page.

Note how the current alt tags on the images actually gets in the way of understanding the content.  Because it breaks the flow.

A better alt tag for this might be:

"line drawing Predicting purchasing patterns, On the left a customer with a market basket of purchased items labeled "Data (customer purchase info), in the Middle a robot representing ML model, the robot is using a set of rules based on "Monthly spending on web site, Products Reviewed, Area of residency, Banks with..., Previous Transactions on web site to make a binary yes/no prediction."

cc: @Alex_Reutter 

 

CoreyS
Community Manager
Community Manager
Status changed to: Acknowledged