Emily Wells
Academic and research departments
Centre for Translation Studies, Leverhulme Doctoral Scholarships Network for AI-enabled Digital Accessibility (ADA).About
My research project
Investigation of AI chatbots or widgets with entity linking for accessibilityAs a Leverhulme Trust Doctoral Scholar, I aim to investigate how AI chatbots could be used to make online instructions more accessible to neurodivergent users by allowing these users to customise the format of instructions according to their individual preferences. I will start with a review of existing literature on web accessibility for neurodivergent users to propose possible theories and methodologies for creating the chatbot widget. The next phase will involve a survey of neurodivergent users to determine which features would be most useful for them, and these features will then be implemented in the chatbot widget. Finally, user feedback on the tool will be used to improve it further. My background is in translation, and I also completed my MA Translation and BA French and Spanish at the University of Surrey.
Supervisors
As a Leverhulme Trust Doctoral Scholar, I aim to investigate how AI chatbots could be used to make online instructions more accessible to neurodivergent users by allowing these users to customise the format of instructions according to their individual preferences. I will start with a review of existing literature on web accessibility for neurodivergent users to propose possible theories and methodologies for creating the chatbot widget. The next phase will involve a survey of neurodivergent users to determine which features would be most useful for them, and these features will then be implemented in the chatbot widget. Finally, user feedback on the tool will be used to improve it further. My background is in translation, and I also completed my MA Translation and BA French and Spanish at the University of Surrey.
Publications
This paper describes the SurreyCTS submission 1 to the BEA 2026 shared task on lexical difficulty prediction, entered in the Open Track. Our approach progressed from base-line multilingual encoders to a hybrid Rem-BERT architecture with extensive feature engineering , combining semantic funnelling, lexical similarity features, attention-derived signals , and language-aware representations. On our internal production validation set, the best single models achieved RMSE 0.8122 (prod-H) and Pearson correlation 0.8968 (prod-G). A weighted ensemble of the five strongest systems, with weights proportional to inverse squared validation RMSE, was submitted as our final entry. On the official shared-task test set, the ensemble achieved RMSE 1.034, 0.945, and 0.861 for Spanish, German, and Chinese respectively, outperforming the open-track base-line in all three settings and placing fifth among open-track teams.