ENDING BIAS IN ANY FORM

CAN COMPUTER VISION AND MACHINE LEARNING MAKE WEB EXPERIENCES MORE ACCESSIBLE?

ENDING BIAS IN ANY FORM

CAN COMPUTER VISION AND MACHINE LEARNING MAKE WEB EXPERIENCES MORE ACCESSIBLE?

Think about how many online forms you fill out each day. Logging into an account. Making a purchase. Setting an appointment. The list goes on.

Though typing into fields and clicking submit over and over can feel tedious, most people accept it as a minor inconvenience of modern life. But what if it was more than inconvenient: What if it were inaccessible?

For blind and visually impaired users, many websites’ forms aren’t usable – even when their code has been vetted by automated accessibility testing tools.

According to an annual study by international web accessibility nonprofit WebAIM, Web Content Accessibility Guidelines 2.0 failures were detected on 96.8% of the top 1 million homepages, and 46% had issues specifically related to forms.

That’s what inspired Umut Gultekin, who works for Swedish web accessibility company Funka, to use the SAS Hackathon as an opportunity to address the problem.

“The users themselves are not disabled,” Gultekin says. “It’s the interface that disables them.”

AN AMBITIOUS VISION

The problem, in a nutshell, is that forms aren’t always as accessible to screen readers as they might appear to be in the code. But to use a website like sighted users, blind and visually impaired users have to be able to trust their screen reader – and by extension, the developers and testers behind each website.

If there’s a failure, those users are simply locked out. They can’t log in. They can’t pay a bill. They can’t schedule an appointment. Worse yet, they might not even know anything went wrong or why.

“Those are nightmarish scenarios for people who use assistive technologies,” Gultekin says. “Basically the users will have no way of knowing what is there without being able to see the form themselves.”

So Gultekin and his colleagues set out to create a novel approach to verifying forms’ accessibility using machine learning and computer vision. Lacking experience with those specific technologies, they sought help through the SAS Hacker’s Hub and teamed up with several AI and ML experts.

“We were more courageous after that,” Gultekin says. “We wanted to experiment with new technologies. That’s why we chose something that was more ambitious, and luckily we were able find team members from Hacker’s Hub with those specific skill sets.”

Funka's entry in the SAS Hackathon, where teams come together globally to turn curiosity into innovation, described the challenge and novel approach to improving web form accessibility.

FIXING THE BIG PICTURE

Funka WebSight, the prototype solution Gultekin’s team developed during the Hackathon, allows anyone to enter a URL, at which point the software takes screenshots of forms on the page and visually identifies viable form elements (for example: input, check box, text area, select, radio, button, CAPTCHA) using computer vision and machine learning.

The solution also finds potential discrepancies between what’s perceivable to the human eye and how the form fields are coded. If discrepancies are found, they are marked as potential accessibility problems. The resulting report shows the discrepancies between what meets the eye and what is there in the form’s code.

New applications of technologies like this are critical to responsible innovation, ensuring that groups of people – even if they are a small part of the overall population – aren’t victims of unintended consequences as technologies are developed.

“Very few organizations have mature accessibility programs and are actually investing in it and go to the trouble of being more inclusive,” Gultekin says. “And that's a bit sad. If more developers paid attention to web standards and best practices in developing web forms, the entire internet would be much more accessible because it’s all the same handful of problems repeating themselves countless times.”

Gultekin and some of the founders of Funka recently established a sister company called Accessibility Cloud to offer better accessibility testing solutions to the global market, where WebSight technology could help create a more inclusive digital world.

“We need to think about all sorts of people with different capabilities,” he says. “Now that we have gotten some serious experience out of the SAS Hackathon, we really want to do more within this field with Accessibility Cloud.”

RECOMMENDED FOR YOU

EDUCATION

BUILDING BLOCKS OF INCLUSIVITY

Accessible coding education resources unleash possibilities for the visually impaired.

→ READ MORE

DATA VIZ

FILLING IN THE BLANKS

Can a writer prove that even data newbies can gain insights from exploring the data in their daily lives?

→ READ MORE

Curious about SAS and the analytics that empower organizations everywhere?

GET TO KNOW US