Modeling Event Salience in Narratives via Barthes' Cardinal Functions

TL;DR

This paper introduces unsupervised methods to estimate event salience in narratives using Barthes' theory and pre-trained language models, improving performance on folktale datasets without requiring annotated data.

Contribution

It proposes novel unsupervised approaches based on Barthes' cardinal functions for assessing event importance in narratives, emphasizing the role of fine-tuning language models on narrative texts.

Findings

Proposed methods outperform baseline approaches.

Fine-tuning language models enhances event salience estimation.

Effective in folktale narrative analysis.

Abstract

Events in a narrative differ in salience: some are more important to the story than others. Estimating event salience is useful for tasks such as story generation, and as a tool for text analysis in narratology and folkloristics. To compute event salience without any annotations, we adopt Barthes' definition of event salience and propose several unsupervised methods that require only a pre-trained language model. Evaluating the proposed methods on folktales with event salience annotation, we show that the proposed methods outperform baseline methods and find fine-tuning a language model on narrative texts is a key factor in improving the proposed methods.