PhD Completion !!!

I am honoured to announce the completion of my PhD in Computer Science at RMIT University.
The thesis can be found in the RMIT Research Repository.

Fairness-Aware Question Answering for Intelligent Assistants

Supervisors:

  • Dr. Damiano Spina
  • Prof. Falk Scholer
  • Prof. Bruce Croft

Abstract

Voice-based question-answering (QA) systems (e.g., Alexa, Siri, ChatGPT) are increasingly relied upon by diverse users, including individuals with visual impairments, low literacy, and the elderly, across domains such as healthcare, workplaces, and hands-free environments.

This shift toward conversational models over traditional search presents new challenges, particularly in how these systems present information. While they offer concise, fact-based answers, they often fail to provide multiple perspectives, reinforcing biases that can shape user perceptions, preferences, and societal attitudes. This limitation is especially critical for stakeholders such as policymakers, healthcare professionals, and researchers, who rely on balanced and diverse information to make informed decisions.

This thesis investigates fairness in multi-perspective rankings for top-heavy voice-based QA systems, focusing on the ranking stage where biases in information exposure can significantly impact users. It introduces methods to estimate fairness across both single and multiple attributes in rankings and empirically differentiates fairness from diversity.

By adopting a multidimensional approach, the research explores techniques for evaluating and mitigating biases in rankings to ensure a more balanced representation of perspectives. Additionally, the thesis incorporates insights from conversational search, information science, psychology, cognitive science, and acoustics, to characterise resources, experimental settings and new research directions for investigating user biases in voice-based QA systems, thereby contributing towards fairness-aware QA in intelligent assistants.


Keywords

Fairness, Diversity, Information Retrieval, Question Answering


Acknowledgements

This research has been conducted on the unceded lands of the Wurundjeri and Boon Wurrung peoples of the eastern Kulin Nation. We pay our respects to their Ancestors and Elders, past and present. This research was partially supported by the Australian Research Council (ARC) under grants DE200100064 and CE200100005.




Enjoy Reading This Article?

Here are some more articles you might like to read next:

  • LLM-Powered Conversational Search (Walert) at RMIT University - Case Studies in Data Science
  • Walert @ RMIT Open Day 2025