歧义输入下的自动形式化：SQL与逻辑形式的精度、覆盖与成本
原标题：Autoformalization for Ambiguous Inputs: Precision, Coverage, and Cost across SQL and Logical Forms
AI 摘要
该论文研究面向歧义输入的自然语言自动形式化问题，在 SQL 与一阶逻辑基准上评估了七种工作流。结果显示，原始模型在 AMBROSIA 和 AmbiQT 上 F1 分别为 53.35% 和 60.85%，而 DRIFT 在 AmP 上以 2.03 倍 token 成本提升 15.98 个 F1 点，LoC 则成本更高但 F1 更低。在 AutoCedar 中，生成前加入结构化解释反而降低 SQL F1 并增加 token 消耗，评审阶段引入 Kimi K3 仅带来约 2 个点的提升。作者结论是，仅靠自动形式化无法保证对歧义的鲁棒性，准确率、歧义处理与推理成本需要联合设计。
正文节选
marginparsep has been altered. topmargin has been altered. marginparwidth has been altered. marginparpush has been altered. The page layout violates the ICML style. Please do not change the page layout, or include packages like geometry, savetrees, or fullpage, which change it for you. We’re not able to reliably undo arbitrary changes to the style. Please remove the offending package(s), or layout-changing commands and try again. Autoformalization for Ambiguous Inputs Precision, Coverage, and Co