用分层树结构解析长篇叙事，提升问答准确率
Storyline Trees: Hierarchical Representations for Long-Form Narratives
- 构建故事线树，从宏观主题到具体事件分层组织叙事
- 在三个长文本问答数据集上，准确率超越强基线模型
- 以场景为单元比章节或通用切分更有效，适合长文理解任务
长篇叙事对长上下文模型构成挑战，因其结构隐含：事件、人物与情节跨数百页交互，缺乏如结构化文档般的显式导航线索。本文提出故事线树（Storyline Trees），一种分层表征方法，将叙事从全局主题和主要情节逐步细化至细粒度事件。首先将章节切分为连续的叙事片段（即场景）作为基本单元；随后通过互补的自上而下与自下而上的过程，在多抽象层级上推导、精炼、聚类并总结故事情节。该表示支持自适应检索，使模型可迭代检查高层叙事结构，并按需获取场景级证据。在三个长上下文叙事问答基准上的实验表明，自适应检索显著优于强基线，包括微调后的长上下文模型与基于代理的分块方法。消融实验验证：以场景为基本单元优于章节或通用分割，且在相同检索预算下性能优势持续存在。
原文摘要 · Abstract (English)
Long-form narratives are challenging for long-context models because their structure is implicit: events, characters, and plotlines interact across hundreds of pages without the explicit cues that guide navigation in structured documents. We address this by constructing storyline trees, hierarchical representations that organize narratives from global themes and major plotlines to fine-grained events. We first segment chapters into contiguous narrative segments, or scenes, and use them as the basic units for tree construction. We then infer storyline trees through complementary top-down and bottom-up procedures that derive, refine, cluster, and summarize storylines at multiple levels of abstraction. We showcase the utility of this representation for question answering: storyline trees enable adaptive retrieval, allowing models to iteratively inspect high-level narrative structure and retrieve scene-level evidence on demand. Experiments on three long-context narrative QA benchmarks show that adaptive retrieval outperforms strong baselines, including post-trained long-context models and agentic chunk-based methods. Ablations confirm that scenes are more effective basic units than chapters or generic segmentation, and that gains persist under matched retrieval budgets
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。