多跳检索中的可预测失败：分数分布置信度评分与弃权（不执行）
Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention
AI总结：
本研究证明多跳检索失败在结构上可预测，提出RegimeAbstain方法，利用检索置信度分数（RCS）实现校准弃权策略，在多个基准上显著降低自信错误答案率（CWAR），并验证了特征机制的领域无关性。
AI中文摘要：
多跳检索失败并非在查询间均匀分布：它们聚集在结构上可预测的子群体中。我们证明了两个结果，将这些结构形式化。第一（CWAR可约性）：当且仅当检索特征携带关于成功的信息时，自信失败减少是可实现的，这一条件由LLM评判流水线满足，但在仅稠密设置中显著较弱，解释了两种机制之间AUC-AC的差距。第二（特征机制互补性）：没有单一的ANN分数特征在所有失败机制中达到最佳预测性能；主导特征在不同数据集间不同（MuSiQue上为查询长度，HoVer上为hop-1集中度），且一个构造性见证对显示每个特征在一种机制中是必要的，在另一种机制中无贡献。我们将这些原则实例化于RegimeAbstain中，它计算一个检索置信度分数（RCS），这是多达九个查询-ANN结构特征的逻辑函数，所有特征无需额外LLM调用即可获得，并使用它实施校准的弃权（不执行）策略。我们定义了自信错误答案率（CWAR）指标，并在三个多跳基准（MuSiQue、2WikiMultiHopQA、HoVer）和两种检索架构（LLM评判和仅稠密）上评估，覆盖五种失败机制，CWAR从14.5%到62.1%。RCS在所有五种条件下对八个置信度基线达到最佳或共同最佳AUC-AC。在MuSiQue（LLM评判）上，RCS在50%覆盖率下将CWAR从39.5%降至20.6%（相对减少47.8%），ECE=0.035。在MuSiQue上训练的模型迁移到2WikiMultiHopQA，仅损失-0.5个百分点AUC，确认了机制特征的领域无关结构。
英文摘要：
Multi-hop retrieval failures are not uniformly distributed across queries: they cluster in structurally predictable subpopulations. We prove two results formalizing this structure. First (CWAR Reducibility): confident-failure reduction is achievable if and only if retrieval features carry mutual information about success, a condition satisfied by LLM-judge pipelines but substantially weaker in dense-only settings, explaining the AUC-AC gap between regimes. Second (Feature Regime Complementarity): no single ANN score feature achieves best predictive performance across all failure regimes; the dominant feature differs between datasets (query length on MuSiQue, hop-1 concentration on HoVer), and a constructive witness pair shows each is necessary in one regime and non-contributory in the other. We instantiate these principles in RegimeAbstain, which computes a Retrieval Confidence Score (RCS), a logistic function of up to nine query-ANN structural features, all available without any additional LLM call, and uses it to implement a calibrated abstention policy. We define the Confident-Wrong-Answer Rate (CWAR) metric and evaluate across three multi-hop benchmarks (MuSiQue, 2WikiMultiHopQA, HoVer) and two retrieval architectures (LLM-judge and dense-only), covering five failure regimes with CWAR from 14.5% to 62.1%. RCS achieves best or co-best AUC-AC in all five conditions against eight confidence baselines. On MuSiQue (LLM-judge), RCS reduces CWAR from 39.5% to 20.6% at 50% coverage (47.8% relative reduction), with ECE=0.035. A model trained on MuSiQue transfers to 2WikiMultiHopQA with only -0.5pp AUC loss, confirming the domain-agnostic structure of regime features.
