Shrome at Touché: Soft-Vote Ensembling and Counter-Causal Augmentation for Causality Extraction
作者: Roham Zendehdel Nobari, Shayan Sooratgar
分类: cs.CL
发布日期: 2026-10-02
备注: 16 pages, 5 figures, 10 tables. Both authors contributed equally. Working notes of Touché at CLEF 2026 (Conference and Labs of the Evaluation Forum), 21-24 September 2026, Jena, Germany
💡 一句话要点
提出软投票集成与反因果增强以解决因果提取问题
🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)
关键词: 因果提取 反因果声明 软投票集成 RoBERTa 自然语言处理 数据增强 机器学习
📋 核心要点
- 现有因果提取方法在处理反因果声明时存在显著不足，容易误判句子的因果关系。
- 本文提出了基于软投票集成和反因果增强的模型，分别针对因果性检测、因果提取和极性标注进行优化。
- 在CCNC测试集上，系统在因果性检测和极性标注上分别达到了F1 0.869和宏F1 0.817，提取得分为0.728，表现优异。
📝 摘要（中文）
Touché 2026扩展了因果提取到反因果声明，即表面上看似因果但实际否定因果关系的新闻句子。现有系统依赖表面线索如“导致”，可能错误地将此类句子判定为因果。本文在Countercausal News Corpus (CCNC)上进行三项子任务：因果性检测、因果与结果提取、极性标注。我们为每个子任务构建了一个模型，检测使用了经过微调的分类器，提取则通过集成三个RoBERTa-large BILOU+CRF标注器的token级得分来实现。极性标注中，针对稀缺的反因果示例，我们利用大型语言模型生成训练句子。最终系统在CCNC测试集上检测F1达到0.869，极性宏F1为0.817，提取得分为0.728，超越所有提交的基线。
🔬 方法详解
问题定义：本文旨在解决因果提取中的反因果声明问题，现有方法依赖表面线索，导致对句子因果性的错误判断。
核心思路：通过构建针对每个子任务的独立模型，结合软投票集成和反因果增强，提升因果提取的准确性和鲁棒性。
技术框架：整体架构包括三个主要模块：因果性检测模型、因果与结果提取模型、极性标注模型。检测模型使用微调的分类器，提取模型通过集成多个标注器的得分，极性标注则利用生成的反因果句子进行训练。
关键创新：最重要的创新在于采用软投票集成方法，而非传统的投票机制，从而提高了提取的准确性。此外，反因果增强策略有效缓解了训练数据稀缺的问题。
关键设计：检测模型使用单一的跨任务规则来消除假阳性，提取模型通过平均token级得分进行解码，极性标注中引入了九种反因果表达模式生成训练句子，确保生成句子通过自动结构检查。
🖼️ 关键图片
📊 实验亮点
在CCNC测试集上，系统在因果性检测任务中达到了F1 0.869，极性标注的宏F1为0.817，提取任务的得分为0.728，显著高于所有提交的基线，展示了模型的优越性能。
🎯 应用场景
该研究在因果关系提取领域具有广泛的应用潜力，尤其是在新闻分析、社交媒体监测和法律文本解析等场景中。通过准确识别因果与反因果关系，能够帮助用户更好地理解信息的真实含义，提升信息处理的智能化水平。
📄 摘要（原文）
Touché 2026 extends causality extraction to counter-causal claims: news sentences whose surface form appears causal but whose meaning denies the causation, as in "It is falsely believed that X caused Y." A system that relies on surface cues such as "caused" or "led to" will accept such a sentence as causal and give it the wrong polarity. On the Countercausal News Corpus (CCNC), the task has three subtasks: deciding whether a sentence is causal (detection), locating its cause and effect spans (extraction), and labeling its polarity as procausal, counter-causal, or uncausal. We build one model per subtask. Detection is a fine-tuned classifier with a single cross-task rule that uses the extracted spans to remove false positives. For extraction, we ensemble three RoBERTa-large BILOU+CRF taggers by averaging their token-level scores before decoding, rather than voting on the spans each tagger produces. For polarity, where labeled counter-causal examples are scarcest, we add training sentences generated by a large language model prompted with nine patterns of counter-causal expression adapted from Hagen et al., keeping only those that pass automatic structural checks. On the held-out CCNC test set, the system reaches F1 0.869 on detection and macro-F1 0.817 on polarity, and in the organizers' final causal-only evaluation of extraction it scores granularity-adjusted F1 0.728, the highest extraction score among all submissions including the organizers' baseline. The development split is used only for component selection and the ablations reported in the paper.