Towards the Automatic Curation of Reading Guides Using Computational Approaches to Aid Literary Exploration

This thesis explores the use of computational methods to partition literary book universes into groups that share the same characteristics. The purpose of this task is to simplify access and personalise the experience of new readers entering the book universe The Cosmere. This is a task which has already been addressed by the fandom through the creation of reading guides. The thesis investigates means to automate this process through computational methods. Employing topic modeling and named entity recognition, the study analyses the textual data – that is, the full-length books - to identify themes and character involvement. A qualitative comparison with a fan-made reading guide is employed to evaluate the results derived from the computational procedures. The findings suggest that named entity recognition has potential in reducing the complexity of entry into complex book universes, while topic modeling failed to deliver satisfactory results. The thesis illustrates how computational methods can be utilised to analyse and provide quantitative insights into literary data. The automation of the reading guides would offer immediate beneftis to readers of the Cosmere, while also demonstrating the possibility of extending this approach to various datasets.