ConStruM:一种结构引导的LLM框架,用于上下文感知的模式匹配
ConStruM: A Structure-Guided LLM Framework for Context-Aware Schema Matching
AI总结:
ConStruM通过结构引导的LLM框架,提供上下文感知的模式匹配,提升匹配准确性。
摘要:
列匹配是在数据整合中协调模式的核心任务。列名和描述对于此任务是有价值的。LLMs可以利用这种自然语言模式元数据。然而,在许多数据集中,正确的匹配需要额外的证据,而不仅仅是列本身。由于给LLM提供整个所需的模式元数据是不现实的,核心挑战是选择和组织最有用的上下文信息。我们提出了ConStruM,一种结构引导的框架,用于预算化的证据打包在模式匹配中。ConStruM构建了一个轻量、可重用的结构,在查询时,它会组装一个小的上下文包,强调最有区分性的证据。ConStruM被设计为一个附加组件:给定由上游匹配器生成的候选目标短名单,它会增强匹配器的最终LLM提示,使其包含结构化的、查询特定的证据,以便最终的选择更加稳固。为此,我们开发了一种上下文树用于预算化的多级上下文检索,以及一个全局相似性超图,该超图通过在线或预计算的组意识区分线索来总结源和目标两侧的相似列组。在真实数据集上的实验表明,ConStruM通过提供和组织正确的上下文证据提高了匹配效果。
英文摘要:
Column matching is a central task in reconciling schemas for data integration. Column names and descriptions are valuable for this task. LLMs can leverage such natural-language schema metadata. However, in many datasets, correct matching requires additional evidence beyond the column itself. Because providing complete schema metadata remains impractical for most LLMs and can obscure the relevant evidence even when it fits, the core challenge becomes to select and organize the most useful contextual information. We present ConStruM, a structure-guided framework for budgeted evidence packing in schema matching. ConStruM constructs a lightweight, reusable structure in which, at query time, it assembles a small context pack emphasizing the most discriminative evidence. ConStruM is designed as an add-on: given a shortlist of candidate targets produced by an upstream matcher, it augments the matcher's final LLM prompt with structured, query-specific evidence so that the final selection is better grounded. For this purpose, we develop a context tree for budgeted multi-level context retrieval and a global similarity hypergraph for grouping and differentiating confusable columns. Across HRS-B and PSID-B, ConStruM achieves a mean combined accuracy of 95.4%, compared with 51.6% for the no-context LLM, and ranks first among all eleven evaluated methods on both benchmarks. On MIMIC--OMOP and Synthea--OMOP, attaching it to our Matchmaker reconstruction produces no statistically significant accuracy change. Together, these results show that ConStruM can deliver large gains when matching depends on context while remaining compatible with a strong matcher in the other tested settings.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。