<aside> 💡

When technology helps and when it excludes.

</aside>

Introduction

Technology is not neutral — it works well for some users and creates friction or exclusion for others. The AccessGuru dataset contains over 3,500 real-world web accessibility violations collected from 448 diverse websites across domains like health, education, government, news, technology, and e-commerce. Each instance is annotated with one of 112 distinct violation types spanning syntactic, semantic, and layout categories defined by WCAG 2.1 guidelines.

As a Datathon participant, your team is tasked with exploring where technology fails, who it excludes, and what barriers exist in digital systems. By analyzing this dataset, you can uncover patterns of inequity, identify hidden assumptions in design, and suggest ways technology can be more inclusive.

This Datathon is your opportunity to dive into real-world accessibility data, explore exclusion points in digital systems, and develop data-driven insights or predictive models that help technology serve everyone more equitably.

Tasks

Your task is to answer one or more of the following questions, or any other question that sparks curiosity in you and your team regarding the dataset:

Machine Learning / Predictive Modeling

Teams interested in ML may attempt tasks such as:

• Predict the type of accessibility violation based on HTML code snippets or metadata.
• Predict violation severity or impact for a new page using page features.
• Cluster websites based on similarity in violation patterns to identify “high-risk” domains.
• Rank domains or pages by likelihood of inaccessible design.

Data Analytics and Data Visualization

The following questions can be attempted by analytics and visualization teams:

• Which domains (health, education, government, etc.) have the highest number of accessibility violations?
• What violation types are most common across sites or domains?
• Are there patterns in violations by violation category (syntactic, semantic, layout)?
• Identify websites/pages with the most severe accessibility issues.