Developing Machine Learning (ML) Models for Contextually Appropriate Alt Text Generation and Evaluation

Alternative text (alt text) plays a vital role in web accessibility, providing descriptions of images for screen reader users, especially those who are blind or visually impaired. However, current machine learning models for automatic alt text generation often lack nuance, context-awareness, and alignment with the expectations and preferences of visually impaired users.

This PhD research aims to explore how machine learning can be used to generate meaningful, context-aware alt text by:

• Building and refining ML models trained on diverse, human-authored alt text data.
• Incorporating human values and user expectations into model training and evaluation.
• Designing evaluation frameworks to assess appropriateness, usefulness, and inclusivity.
• Developing tools to assist content authors in producing better alt text through ML-supported suggestions grounded in best practices.

Areas of focus

• Accessibility
• Machine Learning/AI
• Human-Computer Interaction

How flexible are the above topics?

Flexibility can be considered based on research findings.

Who should apply?

Students with a technical background and skills, and knowledge/experience working with machine learning/AI. Knowledge of accessibility concepts is desirable, but not mandatory.

Brunel offers a number of funding options to research students that help cover the cost of their tuition fees, contribute to living expenses or both. See more information here: https://www.brunel.ac.uk/research/Research-degrees/Research-degree-funding. The UK Government is also offering Doctoral Student Loans for eligible students, and there is some funding available through the Research Councils. Many of our international students benefit from funding provided by their governments or employers. Brunel alumni enjoy tuition fee discounts of 15%.