Hannah Rose Kirk
DPhil Student
Hannah is a 3rd year DPhil student in Social Data Science at the OII. Hannah's research centres on the role of granular and diverse human feedback for aligning large language models.
Thesis title: Hatemoji: The Construction and Classification of Emoji-based Hate Speech
Supervisors: Dr Scott A. Hale, Oxford Internet Institute and Dr Bertie Vidgen, Alan Turing Institute
Hannah said, “When I started this project back in January 2021, researching emoji-based hate seemed niche and few academics had considered emoji as conduits of serious online harm. Six months down the line, emojified hate hit the headlines following the Euros 2020 Final after England’s Black footballers were faced with a storm of racial abuse in emoji form. Since then, my research has been covered by Sky News, ITV, The Telegraph, Bloomberg and featured on radio and podcasts. Despite endless debugging, a global pandemic, moments of imposter syndrome and weekends stolen away by training models, I have always been deeply motivated by the potential of this thesis to diversify the way we think about online hate, so I am honoured that its impact has been recognised by others. I remember reading the words of previous thesis prize winners before I joined the OII and admiring their work. It is truly a source of pride to be able to write my own words today, but this prize goes beyond my own efforts. Special thanks to my brilliant supervisors Dr Scott A. Hale and Dr Bertram Vidgen, and to the Dynabench team, who all wholeheartedly believed in me and my research from the get-go. Everyone in the OII community has been pivotal in this success and the award deserves to be shared among them. To my supervisors, colleagues, fellow students and friends: thank you, I couldn’t have done it without you ❤️.”
In congratulating Hannah, Dr Scott A. Hale said, “I am tremendously proud to have Hannah recognized with this award. From the initial conceptualization through to the design, operationalization, data collection, analysis and writing, Hannah owned every aspect with determination, creativity, and grit in a role closer to that of a post-doc than a MSc student.”
Dr Bertie Vidgen added, “AI tools are now used by nearly all platforms to find and stop online hate — but they have serious weaknesses. Hannah’s research has both presented the field with a new problem (detecting hate expressed with emoji) and offered a promising way of solving it (through data-centric AI techniques). It was a huge pleasure to contribute to this innovative and impactful project.”
Thesis title: Gameplay matters: The relationship between in-game player behaviour, performance, and well-being
Supervisor: Professor Andy Przybylski, Oxford Internet Institute
Thomas said, “I am incredibly honoured to receive the OII thesis Prize. It humbles me that my work was chosen among that of such a brilliant and truly inspiring cohort. This thesis wouldn’t have been possible without the guidance of my excellent supervisors, so I would like to express special thanks to: Prof. Andrew Przybylski for his continued support and thought-provoking mentorship, Dr. Matti Vuorre for his advice on research design and quantitative methodologies, and Dr. Niklas Johannes for inspiring me to choose this topic and providing me with the data. Video games have become the dominant form of entertainment with billions of players across the globe, and I believe that the success of this thesis highlights the need to better understand how digital play influences well-being. I hope my research will contribute to building the robust and empirical evidence base that is required to delineate responsible use and inform effective regulation.”.
Prof Andy Przybylski said, ” For a very long time what we know, or think we know, about human behaviour in video games has been shaped by quantitive social science research grounded entirely in self-report survey methodology. Interest theoretical work, considering individual motivations and player types have been discussed but, in many cases, attempts to translate these perspectives into quantitate models have fallen short. Thomas’s thesis presents an intriguing and promising alternative approach: leveraging objective gameplay data to examine the player-specific factors which might explain how video games shape players’ well-being. He did an excellent job. Our team is keen to see where he takes this programme of work next!”
The Board of Examiners also chose to award four ‘Highly Commended’ prizes this year:
Thesis title: More Than Dots on a Map: Land as Socio-technical Imaginary on the Faroe Islands
Thesis title: Managing Online Rumour Proportions During Offline Protests
Thesis title: The impact of Ofsted reports on demand for state-funded primary schools in England, from 2012-2016
Thesis title: Momentum Gains Attention: Enhancing Deep Time Series Momentum Strategies Using Attention-Based Networks