Mythili K
The 9th edition of the Global Testing Retreat 2024!
About Speaker
Mythili is a seasoned Accessibility Tester at cognizant, driven by a passion for creating inclusive digital environments. She specializes in making web content accessible for individuals with diverse abilities, identifying potential barriers and ensuring compliance with accessibility standards such as WCAG. She is deeply committed to promoting equal access to digital information, contributing significantly to efforts to foster inclusivity. Her technical expertise, combined with her problem-solving skills, allows her to excel under pressure and deliver impactful results across a wide range of projects.
More Speakers
Web accessibility is crucial to ensuring that all users, including those with differently abled people, have equal access to digital content. Challenges arise when accessibility testers encounter web pages in unfamiliar languages. Traditional methods allow only high-level checks, such as validating roles and structural elements, often failing to ensure the accuracy and relevance for all users. This paper proposes an AI-driven framework that automate the validation of web content in languages unknown to testers, using advanced natural language processing (NLP) and machine learning (ML) techniques. The framework focuses on language detection and translation, content validation, and screen reader output verification.
The framework begins with the automatic detection of the webpage’s language. The content is then translated into a language familiar to the tester, allowing for preliminary validation. This translation helps testers understand the structure and content of the page to perform accessibility testing as per WCAG Guidelines.
A crucial component is ensuring that screen readers accurately convey content to visually impaired users. The AI simulates will analyze the output of the screen readers and check whether the content in the webpage is announced properly by the screen readers.
Utilizes advanced language models such as BERT and GPT for processing and analyzing multilingual content.
Employs supervised and unsupervised learning techniques to train models on multilingual datasets, enabling accurate content validation and screen reader output verification.
This paper presents an innovative AI-driven framework for automating the validation of web content in languages unknown to accessibility testers. By leveraging NLP and ML techniques, the solution overcomes the limitations of traditional accessibility testing methods, offering a comprehensive approach to ensuring web accessibility. The framework’s capabilities significantly enhance the validation process, demonstrating the transformative potential of AI in creating a more inclusive digital environment.
