我们是否需要回答那个问题?自然对话中潜在问题的显著性与可回答性
摘要
本文实证研究了自然对话中问题的显著性与可回答性之间的关系,发现一个稳健但低的正相关关系,且弱于独白文本中的相关性,这表明对话结构的可预测性较低。
arXiv:2609.31130v1 Announce Type: new
Abstract: We empirically investigate Question Under Discussion based modelling in naturalistic dialogue by studying whether the salience of generated potential questions predicts their subsequent resolution. Building on Wu et al. (2024), we construct a dataset of 7,124 questions automatically generated from utterances and preceding context from the British National Corpus, and annotated for salience and answerability. We find a robust but low positive correlation between salience and answerability in dialogue, indicating that more salient questions are more likely to be addressed. However, this effect is markedly weaker than in monologic text, suggesting that conversational structure is less predictable. We further observe that structured interactions exhibit stronger alignment between annotators than less organised dialogues.
查看缓存全文
缓存时间: 2026/09/28 09:49
# Do we need to answer that question? Salience and Answerability of Potential Questions in Naturalistic Dialogue Source: [https://arxiv.org/abs/2609.31130](https://arxiv.org/abs/2609.31130) [View PDF](https://arxiv.org/pdf/2609.31130) > Abstract:We empirically investigate Question Under Discussion based modelling in naturalistic dialogue by studying whether the salience of generated potential questions predicts their subsequent resolution\. Building on Wu et al\. \(2024\), we construct a dataset of 7,124 questions automatically generated from utterances and preceding context from the British National Corpus, and annotated for salience and answerability\. We find a robust but low positive correlation between salience and answerability in dialogue, indicating that more salient questions are more likely to be addressed\. However, this effect is markedly weaker than in monologic text, suggesting that conversational structure is less predictable\. We further observe that structured interactions exhibit stronger alignment between annotators than less organised dialogues\. ## Submission history From: Amandine Decker \[[view email](https://arxiv.org/show-email/99eaa4f8/2609.31130)\] \[via CCSD proxy\] **\[v1\]**Fri, 25 Sep 2026 11:23:45 UTC \(1,284 KB\)
相似文章
LLM弃权的两个维度:答案正确性与问题可回答性
本文研究了LLM弃权的两个维度:答案正确性与问题可回答性。研究表明,单一的置信度阈值会混淆这两种失败模式,并提出了一种带有独立预算的三类选择性接受框架。在五个经过指令微调的模型上进行的实验表明,可回答性在内部是可读的,但输出置信度或自我评估难以捕捉。
话题作为社会人口特征的代理:对话上下文如何影响大语言模型回答
本文研究了大语言模型如何因对话上下文而产生不同结果,发现话题而非明确的用户人口特征是导致高风险场景(如薪资建议)中差异的主要驱动因素。
提示鲁棒性取决于任务:LLM评估中客观问题与信念风格问题的比较
本文研究了LLM评估中提示鲁棒性在客观问题和主观问题之间的差异,发现对提示变化的敏感性取决于问题类型、提示变化和模型。
TurnNat:双人对话中轮流发言自然性的自动评估
TurnNat是一种基于似然的框架,用于自动评估双人对话中的轮流发言自然性,它使用在自然对话上训练的因果轮流发言预测模型,通过负对数似然来测量时间异常性。
坍塌后的知识:上下文与措辞塑造语言模型中事实知识的可访问性
该论文提出了知识回答分歧(Knowledge Answer Divergence,KAD),这是一种以事件为条件的评估协议,从语言模型采纳错误纠正的时刻分出探测分支,在八个本地模型检查点和六个托管端点上衡量错误的持续、恢复和弃答情况。研究发现,对话历史和措辞具有显著影响:中性历史分支大幅提高了 Llama 3.2 Instruct 3B 的回答正确率,而提示线索的位置使 Qwen 3.5 4B 采纳错误纠正的比例变化了 27.3 个百分点。