CLADE 论文：RAG 检索到粗粒度证据时，模型会过度猜测而非拒答
CLADE: Category-Level Answering from Deficient Evidence in Retrieval-Augmented Language Models
做 RAG 或知识问答的工程师值得看看这篇。它揭示了一个容易被忽视的坑：检索到的段落粒度偏粗时，模型不会拒答而是瞎猜，错误率约一半。简单加置信度过滤或换更大的模型都没用，论文给出了结合领域本体缓解此问题的路径，以及仍未解决的关键环节。
arXiv 论文《CLADE: Category-Level Answering from Deficient Evidence in Retrieval-Augmented Language Models》提出“可解决的不足”概念：RAG 系统常检索到准确但粒度过粗的段落，例如问某药治疗哪种具体疾病，段落只提到疾病家族。
在 PrimeKG 生物医学问题上，这类“仅家族”段落使模型弃答率下降 13.8 个百分点，但 resulting 猜测约一半是错的。
反事实测试显示，将精确疾病名替换为家族名比直接删除该句使弃答率多降 17.4 个百分点。
论文还发现置信度过滤、换用更大模型、明确指示拒答等简单方法都无法消除该现象，并在 GloBI 生态物种交互数据上复现。用本体论修复时，给定正确家族后候选列表在 98.
5% 的情况下包含真实答案，但字符串匹配只能定位到 19.8%，让模型自行选择家族提升到 30.
0%，端到端恢复率从 10.0% 提升到 26.
8%。
CLADE: Category-Level Answering from Deficient Evidence in Retrieval-Augmented Language Models
Retrieval-augmented systems often retrieve passages that are accurate and relevant but state the answer at a coarser level of granularity than the question requires. For example, a question may ask which specific disease a drug treats (familial hyperlipidemia), while the retrieved passage mentions the broader disease family (hyperlipidemia). Prior work on context sufficiency treats such coarse passages simply as insufficient. We argue that they form a distinct case, which we call resolvable insufficiency: text alone cannot settle the fine-grained answer, but the broader category it mentions maps to a short list of candidate answers in a domain ontology. When language models read these coarse, family-only passages, they exhibit a phenomenon we call Clade (Category-Level Answering from Deficient Evidence): they answer instead of abstaining or searching further. On biomedical questions from PrimeKG, family-only passages reduce abstention by 13.8 points compared to uninformative passages, and the resulting guesses are wrong about half the time. A counterfactual test confirms the cause: replacing an exact disease name with its broader family name lowers abstention by 17.4 points compared to deleting the sentence. Simple fixes like confidence filtering, scaling to a larger model, and explicitly instructing the model to abstain cannot resolve Clade, and this replicates on ecological species interactions in GloBI. Finally, we test whether an ontology can turn coarse passages into exact answers. Given the correct family, an ontology list includes the true answer in 98.5% of cases, but string matching finds that family on the page for only 19.8%. Letting the model choose the family raises this to 30.0% and end-to-end recovery from 10.0% to 26.8%. Overall, our results show when language models should stop guessing and consult an ontology, and highlight which steps of that process remain unsolved.
This report is categorized under AITOP's "Papers" desk. Key entities and technical topics under continuous tracking include RAG, 知识问答, 本体, 拒答.
Source attribution: curated from arXiv cs.AI (@S M Rayeed, Charles V. Stewart, Yao Ma).
