学习数据湖：自适应数据产品发现中的可靠经验
Learning the Lake: Reliable Experience for Adaptive Data Product Discovery
浏览论文内容
中文总结 AI 辅助
本研究提出SafeLake框架，利用演化发现记忆分离操作熟悉度与产品证据，在保持自适应预算下，通过可靠经验决定数据湖搜索收缩时机，节省49.5%-82.7%资产暴露。
中文摘要 AI 辅助
数据产品发现即使在工作负载重复访问相关产品和区域时，也会搜索整个数据湖。重复允许收缩搜索，但相似性不能证明路线的合理性，因为一个被遗漏的资产会使合取产品失效。我们研究服务经验何时能安全地减少这项工作。演化发现记忆在固定区域索引之上记录带源标签的查询-产品-区域证据。SafeLake将操作熟悉度（决定搜索多少）与独立校准的产品证据（决定搜索哪里）分开。固定探针比较在SafeLake和仅熟悉度之间保持自适应预算恒定。在TAT-QA上，产品引导将产品召回率提高了0.072；ConvFinQA显示没有解决的地图增益，而HybridQA的敏感性在完整R@100中偏向仅熟悉度。仅轨迹、缺失和虚假反馈暴露了地图引导的边界，而范围审计协议无法认证来源。在冻结的转导协议下，在干净确认的反馈流中，形式化控制器节省了49.5%至82.7%的累积资产暴露。经验决定何时收缩；可靠证据决定在哪里收缩。
英文摘要
Data-product discovery searches a full lake even when workloads revisit related products and regions. Repetition permits contracted search, but similarity cannot justify a route because one omitted asset invalidates a conjunctive product. We study when serving experience can safely reduce this work. Evolving Discovery Memory records source-labelled query--product--region evidence above a fixed regional index. SafeLake separates operational familiarity, which determines how much to search, from independently calibrated product evidence, which determines where to search. The fixed-probe comparison holds the adaptive budget constant between SafeLake and Familiarity-only. On TAT-QA, product steering raises Product Recall by 0.072; ConvFinQA shows no resolved map gain, while the HybridQA sensitivity favors Familiarity-only in Full R@100. Trace-only, missing, and false feedback expose boundaries on map steering, while scope-audit agreement cannot certify the source. Across clean confirmed-feedback streams under the frozen transductive protocol, the formal controller saves 49.5--82.7% of cumulative asset exposure. Experience determines when to contract; reliable evidence determines where to contract.
发表机构
- Hong Kong Baptist University（香港浸会大学）
- The University of Tokyo（东京大学）
- Tokyo University of Science（东京理科大学）
- ShanghaiTech University（上海科技大学）
机构由 AI 辅助整理，请以论文原文为准。
