S^2tory: Story Spine Distillation for Movie Script Summarization

2026-05-04

AI Summary

This work addresses the challenge of summarizing screenplays with nonlinear narrative structures, which often confound conventional summarization methods by obscuring core plot elements. To tackle this, the study introduces a novel framework grounded in narratological theory, leveraging character development trajectories to identify essential plot nuclei. A Narrative Expert Agent performs theory-constrained reasoning to guide the selection of pivotal events, while knowledge distillation transfers this structural insight to a lightweight model that prioritizes the narrative backbone. The approach employs a two-stage generation mechanism and achieves state-of-the-art semantic fidelity on the MovieSum dataset at a compression ratio of approximately 3.5×. Moreover, it demonstrates strong zero-shot transfer performance on BookSum, and human evaluations confirm the efficacy of narratological principles in modeling complex, nonlinear storytelling.

Abstract

Movie scripts pose a fundamental challenge for automatic summarization due to their non-linear, cross-cut narrative structure, which makes surface-level saliency methods ineffective at preserving core story progression. To address this, we introduce S^2tory (Story Spine Distillation), a narratology-grounded framework that leverages character development trajectories to identify plot nuclei, the essential events that drive the narrative forward, while filtering out peripheral satellite events that merely enrich atmosphere or emotion. Our Narrative Expert Agent (NEAgent) performs theory-constrained reasoning, whose distilled knowledge conditions a small model to identify plot nuclei. Another model then uses these plot nuclei to generate the summary. Experiments on the MovieSum dataset demonstrate state-of-the-art semantic fidelity at approximately 3.5x compression, and zero-shot evaluation on BookSum confirms strong out-of-domain generalization. Human evaluation further validates that narratological theory provides an indispensable foundation for modeling complex, non-linear narratives.

Problem

• movie script summarization
• non-linear narrative
• plot nuclei
• story progression
• automatic summarization

Innovation

• story spine distillation
• plot nuclei
• narrative expert agent
• character trajectory
• non-linear narrative summarization