The text-to-SQL task aims to convert natural language questions into corresponding SQL queries based on a given database schema. Previous models that rely on graph neural networks often struggle to accurately capture the complex grammatical relationships present in these questions, leading to poor performance when generating queries for longer requests. To address these challenges, we propose RGISQL, which integrates refined grammatical information extracted from the question and employs segmentation processing to effectively manage long queries. Additionally, RGISQL minimizes the complexity of edge embeddings by reducing the coupling within graph neural networks. By utilizing grammatical dependency trees, RGISQL is better equipped to capture the inherent structure and grammatical rules of questions. This refined grammatical information offers additional contextual and semantic cues for the model, thereby enhancing both its generalizability and interpretability. Furthermore, we dynamically assess the importance of different edges based on the graph structure, which helps reduce the coupling of edge embeddings and further improves the model’s performance. Multiple sets of experiments conducted on the Spider and Spider-Syn datasets demonstrate that RGISQL outperforms other baselines, achieving the best results in both datasets.

In text-to-SQL, generating accurate SQL queries is crucial. Existing methods that rely on Graph Neural Networks (GNN) still fall short in capturing complex semantic and relational information. We propose TFFSQL, which obtains the textual information of natural language questions as node features and integrates them with the original node features. It also adopts segmentation processing to handle long queries, generates vectors for each segment, and merges them with the node structural features, effectively capturing the semantic information of natural language queries and improving the efficiency and accuracy of the model in processing long queries. An cross-graph attention is introduced in the Relational Graph Attention Network (RGAT). Through the cross-graph attention, the model can dynamically adjust attention weights between the original graph and the line graph, thereby capturing rich relational information. Additionally, TFFSQL also combines recursive decoding with the ASTormer decoder for decoding. Experimental results show that our method outperforms existing GNN-based methods on the Spider dataset, significantly improving the accuracy of SQL query generation.