Large Language Models (LLMs) have demonstrated remarkable capabilities in understanding and processing complex text and have recently shown promise in tasks of extracting information from unstructured and semi-structured documents and records. This can be used to summarise and analyse literature, which is beneficial to researchers as it reduces cognitive load by allowing them to judge the relevance of theses quickly, making common themes and methods stand out.

The aim of this project is to address the lack of systematic tools available for analysis and comparison of PhDs and professional doctorates, within the domain of Social Work, by developing an LLM-based automated semantic analysis tool to aid in research tasks, reducing the amount of time-consuming and costly work, increasing the ability to identify trends, gaps and patterns in the field, and enabling continual monitoring of the research outputs. The system will analyse the full text of doctoral theses to identify key characteristics such as theme, research method, sample size, applied or theoretical orientation, and thesis length. A key challenge is achieving reliable and consistent categorisations across diverse thesis structures and writing styles using characteristics that vary in explicitness.