Improving topic modeling for literary studies: a hybrid model combined with Word2Vec visualization in the case of Robinson Crusoe.

This article examines the application of latent Dirichlet allocation (LDA), a popular topic modeling technique, to the literary analysis of Daniel Defoe's *Robinson Crusoe*, addressing challenges posed by the complexity and length of literary texts. It proposes a hybrid approach combining LDA with Word2Vec, a word vector representation method, to enhance the interpretability and visualization of topic clusters by mapping semantically related words in a three-dimensional space. The study validates this method through comparative analyses of various children's adaptations of *Robinson Crusoe* and other Defoe novels, demonstrating its ability to detect nuanced thematic differences, such as shifts in introspective content. The findings suggest that while LDA was originally designed for short, straightforward texts, its integration with semantic visualization tools like Word2Vec holds promising potential for nuanced literary research.