PhD in Leveraging Natural Language Processing for Data-Driven Agent-Based Modelling of Online Cultural Dynamics

About the Project

Project details:

Understanding rapidly changing digital environments requires both large-scale empirical data and mechanistic models. Computational Social Science (CSS) provides powerful tools to describe the spread of information in online environments — e.g. topic modelling, word embeddings, network analysis — but these studies often remain descriptive, offering limited insight into the underlying generative processes. Complex Systems Modelling (CSM), especially Agent-Based Models (ABMs), offers a formal framework to simulate information diffusion, yet these models frequently rely on highly simplified assumptions and oversimplified variables which fail to capture the full dynamics of information in online environments. 

This project will develop a new integrated methodological toolkit that unites data-rich CSS approaches with theory-driven CSM, enabling researchers to move beyond either purely descriptive analyses or purely stylised modelling. The goal is to make ABMs empirically grounded and interpretable using real digital trace data, and to give CSS studies explicit generative mechanisms for observed online cultural dynamics. The student will first design pipelines to capture and structure online discourse at scale, using state-of-the-art Natural Language Processing (NLP) — e.g. large language model (LLM)-enhanced topic modelling, semantic networks, and moral/value framing detection — combined with network science to represent cultural spaces. This will provide rich, multidimensional cultural data beyond simple frequency counts or hand-coded categories. These empirical data will then inform the development of next-generation ABMs of cultural evolution. Instead of using abstract “opinions” or “traits”, models will be parameterised with features extracted from real-world digital conversations: distributions of semantic content, cultural distance metrics, and diffusion pathways measured on platforms such as X/Twitter, YouTube, or Reddit.  

The project will also explore techniques for “closing the loop” between data and models, including likelihood-free inference (e.g. Approximate Bayesian Computation) and simulationbased calibration, to ensure the ABMs remain predictive and falsifiable rather than narrativeonly. The case study focus will be human-GenAI cultural coevolution — a timely domain where both large-scale data and new theory are urgently needed. We will analyse how GenAI shapes cultural variation (e.g. convergence or fragmentation of discourse), inheritance (reuse and remixing of AI-generated content), and selection (amplification of particular values or narratives). These insights will be formalised into models that simulate long-term feedback loops between humans and GenAI, complementing descriptive studies and providing a deeper causal understanding of AI-driven cultural change. The student will be co-supervised by Dr Chico Camargo, a computer scientist and Turing Fellow specialising in Natural Language Processing and CSS methods, and Prof Alex Mesoudi, an international leader in cultural evolution and specialist in Agent-Based Models. This interdisciplinary mentorship will give the student a unique skill set spanning complex modelling, large-scale data analysis, and AI research. Training will cover advanced NLP pipelines, network and complexity science, and agent-based simulation frameworks, preparing the student for cutting-edge AI and data science careers. By building an end-to-end workflow that links observational data to explanatory simulation, this project directly advances EPSRC’s Artificial Intelligence and ICT themes. Its outputs will help UK researchers and policymakers better understand and anticipate AI-driven cultural dynamics, informing safe and beneficial AI deployment. 

Please direct project specific enquiries to: Dr Chico Camargo Please ensure you read the entry requirements for the potential programme you are applying for. To Apply for this project please click the 'Apply' button

Funding Comment per 3.5 years  

Location 

Streatham Campus  

Advert information