Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MARCUS: An Event-Centric NLP Pipeline that generates Character Arcs from Narratives

Sriharsh Bhyravajjula, Ujwal Narayan, Manish Shrivastava

TL;DR

Problem: quantify character journeys in narratives by modeling events and evolving inter-character relations. Approach: MARCUS is an event-centric NLP pipeline that extracts events, participant roles, sentiment, and implied emotion to produce relation arcs, which are aggregated into character arcs. Findings: evaluated on Harry Potter and The Lord of the Rings, with human surveys showing meaningful alignment between arcs and plots and general agreement on event-to-arc mappings. Significance: provides a tangible, graph-based representation of character development with potential applications in narrative analysis, recommendations, and NLG for fiction.

Abstract

Character arcs are important theoretical devices employed in literary studies to understand character journeys, identify tropes across literary genres, and establish similarities between narratives. This work addresses the novel task of computationally generating event-centric, relation-based character arcs from narratives. Providing a quantitative representation for arcs brings tangibility to a theoretical concept and paves the way for subsequent applications. We present MARCUS (Modelling Arcs for Understanding Stories), an NLP pipeline that extracts events, participant characters, implied emotion, and sentiment to model inter-character relations. MARCUS tracks and aggregates these relations across the narrative to generate character arcs as graphical plots. We generate character arcs from two extended fantasy series, Harry Potter and Lord of the Rings. We evaluate our approach before outlining existing challenges, suggesting applications of our pipeline, and discussing future work.

MARCUS: An Event-Centric NLP Pipeline that generates Character Arcs from Narratives

TL;DR

Problem: quantify character journeys in narratives by modeling events and evolving inter-character relations. Approach: MARCUS is an event-centric NLP pipeline that extracts events, participant roles, sentiment, and implied emotion to produce relation arcs, which are aggregated into character arcs. Findings: evaluated on Harry Potter and The Lord of the Rings, with human surveys showing meaningful alignment between arcs and plots and general agreement on event-to-arc mappings. Significance: provides a tangible, graph-based representation of character development with potential applications in narrative analysis, recommendations, and NLG for fiction.

Abstract

Character arcs are important theoretical devices employed in literary studies to understand character journeys, identify tropes across literary genres, and establish similarities between narratives. This work addresses the novel task of computationally generating event-centric, relation-based character arcs from narratives. Providing a quantitative representation for arcs brings tangibility to a theoretical concept and paves the way for subsequent applications. We present MARCUS (Modelling Arcs for Understanding Stories), an NLP pipeline that extracts events, participant characters, implied emotion, and sentiment to model inter-character relations. MARCUS tracks and aggregates these relations across the narrative to generate character arcs as graphical plots. We generate character arcs from two extended fantasy series, Harry Potter and Lord of the Rings. We evaluate our approach before outlining existing challenges, suggesting applications of our pipeline, and discussing future work.
Paper Structure (20 sections, 2 equations, 3 figures, 2 tables)

This paper contains 20 sections, 2 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Key Concepts
  3. Events
  4. Circumstance
  5. Relation Arcs and Character Arcs
  6. MARCUS
  7. Dataset
  8. BookNLP
  9. Event Extraction
  10. Participant Extraction
  11. Circumstance
  12. Sentiment Identification
  13. Emotion Identification
  14. Relation and Character Arcs
  15. Evaluation
  16. ...and 5 more sections

Figures (3)

  • Figure 1: Relation-Based Character Arc for Frodo Baggins across The Lord of the Rings trilogy. The blue line represents Frodo the actor, and the red line represents Frodo the experiencer. A: Frodo is stabbed by the poisoned blade of the Nazgul at Weathertop (valley). B: Frodo reunites with loved ones at Rivendell (peak). C: Frodo is in grief after the wizard Gandalf falls to a Balrog (valley). D: Frodo is attacked by Shelob in her lair (valley). E: Frodo succeeds in his quest and returns home to The Shire (peak).
  • Figure 2: MARCUS (Modeling Arcs for Understanding Stories), an NLP pipeline that plots a character's arc as their quantitative interaction with circumstance as both actor and experiencer, represented by the proxy amalgamation of their event-centric relations across the narrative.
  • Figure 3: (a) Character Arc for Hermione, No Rolling Function Applied; (b) Character Arc for Hermione, with a savgol filter of window size 1/10th of her event sequence length, fitted with a third degree polynomial.