Social Networks Analysis in Discovering the Narrative Structure of Literary Fiction

TL;DR

This paper explores using social network analysis to understand narrative structures in literary fiction, comparing automated NLP methods with human responses to address the subjectivity of reader interpretation.

Contribution

It introduces a framework for analyzing social interactions in fiction through network theory and evaluates the effectiveness of NLP methods against human perceptions.

Findings

Automated NLP methods can detect social interactions but have limitations.

Interaction networks do not always align with narrative climax.

Subjectivity significantly influences network analysis outcomes.

Abstract

In our paper we would like to make a cross-disciplinary leap and use the tools of network theory to understand and explore narrative structure in literary fiction, an approach that is still underestimated. However, the systems in fiction are sensitive to readers subjectivity and attention must to be paid to different methods of extracting networks. The project aims at investigating into different ways social interactions are read in texts by comparing networks produced by automated algorithms-natural language processing with those created by surveying more subjective human responses. Conversation networks from fiction have been already extracted by scientists, but the more general framework surrounding these interactions was missing. We propose several NLP methods for detecting interactions and test them against a range of human perceptions. In doing so, we have pointed to some limitations of using network analysis to test literary theory, e.g. interaction, which corresponds to the plot, does not form climax.