Exploring how literary authors influence each other and how intertextual connections can be found among them is a complex problem that is handled manually by experts. This research field, intertextuality, seeks to understand how writers relate to each other and how the works of one writer have reflected on the work of another. This analysis usually provides insight into the historical, cultural, biographical, and literary contexts that shape literary works. From a computational point of view, the approaches related to this problem have mainly dealt with problems of style classification, known as authorship attribution, but none have so far dealt with the problem of intertextuality. This paper proposes a novel approach based on contrastive learning to generate authorship embeddings that encapsulate the stylistic signatures of narrative writers and that can be projected into graphical representations to discern similarities between a dataset of literary writers. The embeddings generated with this approach aim to represent authorship styles are created and evaluated using a dataset of books from Project Gutemberg. During the evaluation, we perform a book and chunk-level evaluation, showing good performance in both cases. Finally, we present different graphical representations and provide a deep analysis of the relations that arise between the writers.