PhD Studentship: Multimodal Large Language Models for Audio-aware Argument Extraction

Project title: Multimodal Large Language Models for Audio-aware Argument Extraction  

Supervisory Team: Dr Rafael Mestre, Dr Stuart E. Middleton, Dr Matt Ryan

Project description

For any democratic society, access to news and engaging in online debate are critical for the public to shape opinions and make informed decisions. Modern consumption has moved away from traditional broadcaster mediated content and towards unmoderated social media. In these unmoderated environments, accurate and balanced arguments are directly challenged by the proliferation of misinformation, toxicity, and polarised viewpoints. Use of multimodal Natural Language Processing (NLP) can scale up our ability to extract arguments from online content, both to see key issues and evidence and helping online moderators facilitate healthier online discourse.

The last two years have seen dramatic advances in Large Language Models (LLMs), both for text and multi-modal NLP applications. This PhD will explore how the latest deep learning-based LLM’s can use methods such as chain-of-thought prompting to extract argument components, such as premise, claim and cited evidence, from politician and influencer online video content. You will build a novel LLM architecture that can also directly exploit audio features, such as emotional rhetoric and speech style, so models attend to implicit as well as explicit cues in target videos. A range of attention algorithms will be explored to achieve this, as well as multi-task training using both audio and text task objectives.

This PhD work will be aligned with the Rebooting Democracy project to give opportunities for collaboration with a large inter-disciplinary team and maximise impact.

Prospective candidates are invited to apply soon, as we will be reviewing applications on a rolling basis. Successful candidates would have a strong background in Computer Sciences, Engineering, Maths or Physics and preference would be given to those with good understanding of Deep Learning models and demonstrable interest in interdisciplinary research. You will join the School of Electronics and Computer Science, ranked 1st in the UK for Electrical and Electronic Engineering within the University of Southampton, ranked top 1% of universities worldwide. We’ll also support the development of your future career and give you opportunities including teaching assistantships, professional networking via leading organizations like Alan Turing Institute, £31M RAI UK Hub, and access to Future Worlds to explore commercialization of your research. 

The University of Southampton and its School of Electronics & Computer Science promote equality, diversity, and inclusivity, as demonstrated by our Athena SWAN awards. We also offer benefits for employees' well-being and work-life balance and are committed to sustainability with the Platinum EcoAward. We deeply encourage submissions from applicants from underrepresented communities and welcome applicants that value diversity and are willing to play their part in supporting inclusivity.

Entry Requirements

A very good undergraduate degree (at least a UK 2:1 honours degree, or its international equivalent).

Closing date: 31 January 2024.

Funding: For UK students, Tuition Fees and stipend of £18,622 p.a. for up to 3.5 years.

How To Apply

Apply online: Search for a Postgraduate Programme of Study (soton.ac.uk). Select programme type (Research), 2024/25, Faculty of Physical Sciences and Engineering, next page select “PhD Computer Science (Full time)”. In Section 2 of the application form you should insert the name of the supervisor Rafael Mestre

Applications should include:

• Research Proposal
• Curriculum Vitae
• Two reference letters
• Degree Transcripts/Certificates to date

For further information please contact: feps-pgr-apply@soton.ac.uk

Advert information