RAG-RST: Retrieved Exemplars as In-Context Discourse Schema for RST Parsing

📄 arXiv: 2610.22150v1 📥 PDF

作者: Mohammadreza Sediqin, Shlomo Engelson Argamon

分类: cs.AI

发布日期: 2026-10-15


💡 一句话要点

提出RAG-RST框架以提升RST解析性能

🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)

关键词: 修辞结构理论 话语解析 增强检索 自然语言处理 机器学习

📋 核心要点

  1. 现有的RST解析方法在核性和关系标签的预测上存在显著不足,导致解析性能不理想。
  2. 本文提出RAG-RST框架,通过检索增强的方式提供话语结构示例,以提升解析准确性。
  3. 实验结果表明,RAG-RST在核性和关系的F1分数上分别提升至0.846和0.788,超越了现有的RST解析器。

📝 摘要(中文)

在修辞结构理论(RST)中,话语树的每个边缘都携带核性标签和功能类别关系。现有方法在预测这些标签时表现不佳,强大的语言模型在仅依赖目标EDU对时,核性和关系的F1分数分别为0.794和0.738。本文提出RAG-RST,一个增强检索的框架,通过上下文示例提供正式话语结构,而非通过参数更新,核性和关系的F1分数分别提升至0.846和0.788。比较了预计算和动态检索两种策略,动态策略在各项指标上表现最佳,表明可以通过示例而非训练参数来提供形式语言结构。

🔬 方法详解

问题定义:本文旨在解决现有RST解析方法在核性和关系标签预测上的不足,现有强大语言模型在仅依赖目标EDU对时,表现不佳。

核心思路:RAG-RST框架通过检索增强的方式,提供上下文示例作为正式话语结构,而不是通过参数更新来实现,从而提升解析性能。

技术框架:该框架包含两个主要配置:预计算策略和动态策略。预计算策略在解析前对所有EDU对进行标注,而动态策略在每个减少步骤中进行检索,使得生成器接收到的示例依赖于解析器已做的减少决策。

关键创新:RAG-RST的核心创新在于通过示例提供形式语言结构,而不是将其训练到模型参数中。这一方法在性能上显著优于现有的RST解析器。

关键设计:在设计中,检索的示例数量对性能有界限,超过50个示例后效用会饱和并下降,同时示例的有效性还依赖于其结构位置。

🖼️ 关键图片

fig_0
fig_1

📊 实验亮点

实验结果显示,RAG-RST框架在核性和关系的F1分数上分别达到了0.846和0.788,相较于现有方法有显著提升。动态策略在各项指标上表现最佳,表明其在实际应用中的有效性和优势。

🎯 应用场景

RAG-RST框架在自然语言处理领域具有广泛的应用潜力,尤其是在文本理解、信息提取和对话系统中。通过提升RST解析的准确性,该方法能够改善机器对复杂文本的理解能力,进而推动智能助手和自动摘要等应用的发展。

📄 摘要(原文)

In Rhetorical Structure Theory (RST), each edge of a discourse tree carries a nuclearity label, marking which span holds the central content, and a relation drawn from a closed inventory of functional categories. Neither label is a surface property: the same two clauses can be Elaboration or Background depending on rhetorical intent, and the boundary between them is fixed by annotation convention rather than by lexical cues. Predicting these labels therefore requires a formal discourse schema, and exposure to discourse-rich pretraining text does not supply it: given only a target EDU pair, a strong LLM recovers nuclearity at 0.794 F1 and relation at 0.738 F1. We propose RAG-RST, a retrieval-augmented framework that supplies this schema as in-context demonstration rather than through parameter updates, raising nuclearity to 0.846 F1 and relation to 0.788 F1 with no training. Two configurations are compared: a precomputed strategy that labels all EDU pairs before parsing, and a dynamic strategy that retrieves at each reduce step, so that the exemplars reaching the generator depend on the reductions the parser has already committed to. Both outperform existing RST parsers across span, nuclearity, relation, and full tree metrics, with the dynamic variant consistently strongest. Two properties of the retrieved schema emerge: its utility is bounded in number, saturating and then declining beyond fifty exemplars, and it depends on structural position. Both indicate that a formal linguistic schema can be supplied to a foundation model as a demonstration rather than trained into its parameters.