Schema linking, the task of identifying relevant database schema elements (tables and columns) for natural language queries, is a critical component in database-driven natural language interfaces. While existing approaches rely on question decomposition to handle complex queries, they often suffer from error propagation and low precision. In this paper, we propose a novel schema linking framework enhanced by self-verification (SV) and value hints (VHs) that significantly improves both precision and recall. Our approach introduces two key components: (1) self-verification (SV), an iterative refinement mechanism that validates and corrects initial predictions through explicit verification prompts, and (2) value hints (VHs), which explicitly guide the model to recognize database values mentioned in queries. We conduct comprehensive experiments on two benchmark datasets, Spider and BIRD, using two language models of 4B and 80B parameters. Our results demonstrate that SV + VH consistently improves performance across datasets, models, and method configurations, outperforming both decomposition-based approaches and compute-matched alternatives such as self-consistency under equivalent inference budgets.

In the era of big data and artificial intelligence, effectively integrating and utilizing scattered, heterogeneous multi source data has become crucial for unlocking data value and driving intelligent decision making. Traditional methods, which mostly rely on lexical similarity or manually defined rules, struggle to handle real world scenarios with large scale data and deep implicit semantics. To address these issues, this paper proposes a three level cascaded intelligent matching framework. First, the source and target schemas are abstracted into attribute graphs, and a dual channel embedding approach based on pre trained language models and graph neural networks is adopted to capture both lexical semantics and internal structural dependencies. Innovatively, a domain knowledge graph is introduced as an external semantic hub. Through graph embedding and relational reasoning, scattered attribute terms are mapped into a unified conceptual space, enabling explicit completion and calibration of the original semantics, thereby generating a knowledge enhanced unified semantic representation. Finally, the enhanced schema graphs are constructed into a weighted association graph, transforming the optimal schema mapping problem into a combinatorial optimization task of finding the maximum weight clique in the association graph. Experimental results show that the proposed method significantly outperforms traditional baselines in precision, recall, and F1 score.