弥合数据集搜索中的搜索意图与表格数据
原标题：Bridging Search Intent and Tabular Data in Dataset Search
AI 摘要
该论文提出两个检索组件以支持基于自然语言意图的数据集搜索：Hypothetical Schema Embeddings (HySE) 利用语言模型将用户分析任务描述转化为任务对齐的假设表结构，用于稠密检索；FlexPrune 将自然语言约束翻译为可定制元数据索引上的 SQL 谓词，实现任意元数据属性过滤。作者构建了包含 6,503 张表的 KaggleDS 语料，实验显示 HySE 在召回率上优于关键词和嵌入基线，并在 NQ-Tables 上使 Recall@10 提升 6.0 个百分点；HySE 与 FlexPrune 结合后，相比稠密检索加标签过滤再提升 6.0 个百分点。
正文节选
Bridging Search Intent and Tabular Data in Dataset Search Abstract Dataset search remains a bottleneck in data analysis workflows, as existing systems largely rely on keyword matching or shallow retrieval that fail to capture user intent. To facilitate intent-driven search, we present two retrieval components to equip multi-faceted dataset search interfaces through natural language. First, we propose Hypothetical Schema Embeddings (HySE) that uses a language model to generate task-aligned hypoth