发表机构
Tencent SSV; School of Future Cities, University of Science and Technology Beijing（腾讯SSV; 北京科技大学未来城市学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
Socrates-RAG提出基于前提的主动检索策略，通过选择未解决前提并优化查询，在开放语料库问答中对抗证据投毒，将安全准确率从79.2%提升至93.8%。
AI 中文摘要
检索增强生成（RAG）的防御机制通常决定如何过滤或聚合一个固定的检索集合。然而，在开放语料库问答中，决定性的证据可能不在初始上下文中，但可以通过检索获得，这使得下一次查询成为可靠性问题的一部分。我们引入了Socrates-RAG，一种基于前提的主动检索策略，该策略表示相互竞争的答案，选择一个其解决能够区分这些答案的未解决前提，并在回答或弃权（不执行）之前利用新获取的证据来优化后续查询。我们形式化了由此产生的有限预算证据状态，并给出了相对于重复或主题查询策略的条件性救援保证。我们将Socrates-RAG与一个匹配的对照组进行评估，在该对照组中，相同的骨干模型生成普通的面向相关性的搜索查询；两种策略共享初始证据、确定性检索器、两次查询/前三项预算、答案提示和无标签证据链发布规则。在不相交的48个世界反事实评估中，基于前提的查询将安全准确率从79.2%提升至93.8%，其中8胜1负39平（双侧精确p=.0391）。决定性证据召回率以相同幅度提高，而不安全答案从1个降至0个。两种策略都解决了所有24个单跳案例；收益集中在两跳案例中，其中Socrates-RAG将新解决的前提替换到其第二次查询中。这项受控研究隔离了基于前提获取的特定益处，而不声称在开放网络上具有普遍鲁棒性。
英文摘要
Retrieval-augmented generation (RAG) defenses typically decide how to filter or aggregate a fixed retrieved set. In open-corpus question answering, however, decisive evidence may be absent from the initial context but retrievable, making the next query part of the reliability problem. We introduce Socrates-RAG, a premise-directed active retrieval policy that represents competing answers, selects an unresolved premise whose resolution would discriminate them, and uses newly acquired evidence to refine a subsequent query before answering or abstaining. We formalize the resulting finite-budget evidence state and give a conditional rescue guarantee relative to repeated or topical-query policies. We evaluate Socrates-RAG against a matched control in which the same backbone generates ordinary relevance-oriented search queries; both policies share the initial evidence, deterministic retriever, two-query/top-three budget, answer prompt, and label-free evidence-chain release rule. On a disjoint 48-world counterfactual evaluation, premise-directed inquiry raises safe accuracy from 79.2% to 93.8%, with 8 wins, 1 loss, and 39 ties (two-sided exact p=.0391). Decisive-evidence recall improves by the same margin, while unsafe answers fall from one to zero. Both policies solve all 24 one-hop cases; the gain is concentrated in two-hop cases, where Socrates-RAG substitutes a newly resolved premise into its second query. This controlled study isolates a specific benefit of premise-directed acquisition without claiming general robustness on the open Web.
