semantic parsing

Existing text-to-SQL methods exhibit significant performance variance in cross-database generalization, primarily due to the lack of systematic alignment between domain semantics embedded in natural language queries and structural patterns in database schemas, compounded by inefficient, non-generalizable manual prompt engineering for domain knowledge injection. Method: We propose a structured-domain-knowledge-based multi-database text-to-SQL framework that explicitly models domain knowledge as retrievable, structured statements; employs lightweight substring matching for database-adaptive retrieval; and seamlessly integrates retrieved knowledge into the LLM’s reasoning pipeline—eliminating reliance on handcrafted prompts. Contribution/Results: Evaluated across 11 real-world databases and 5 open-source and commercial LLMs, our approach achieves substantial gains in SQL execution accuracy over strong baselines. It is the first to enable plug-and-play cross-database transfer of domain knowledge, markedly improving model robustness in understanding semantic correspondences between domain vocabulary and schema elements.

This work addresses the challenge of requiring fully formalized inputs during data ingestion—a bottleneck arising from the gap between natural language and formal semantic models—by proposing a “Semantic Ladder” framework. This approach enables a progressive transformation from textual fragments to higher-order logical models through modular semantic units. By integrating natural language, ontological models, and vector embeddings, the framework constructs a multi-layered, traceable semantic representation system. The resulting semantic knowledge infrastructure is both extensible and interoperable, substantially reducing the burden of semantic parsing while supporting the incremental integration and reasoning over heterogeneous knowledge sources. This provides AI systems with a unified yet flexible semantic foundation.

E-commerce platforms generate vast volumes of unstructured product data, exacerbating challenges in information retrieval and cross-system interoperability. To address this, we propose a large language model (LLM)-based multi-agent semantic mapping framework that enables automated, interpretable transformation from relational databases to knowledge graphs. The framework adopts Schema.org as a unified semantic ontology and orchestrates specialized agents to perform table- and column-level semantic annotation, schema alignment, and RDF triple generation—substantially reducing manual knowledge graph construction effort. Empirical evaluation across multiple business scenarios achieves over 90% semantic mapping accuracy, effectively supporting heterogeneous data integration and sharing. Our key contribution lies in the tight integration of LLM-driven semantic understanding with multi-agent collaboration, enabling—for the first time—the end-to-end, high-precision, and scalable automatic construction of product-domain knowledge graphs.

Enterprise multi-source heterogeneous databases engender data silos and impede semantic interoperability. To address this, we propose a multi-agent collaborative semantic mapping framework wherein large language models serve as semantic agents to automatically align relational database tables and columns with the Schema.org standard ontology, thereby constructing a unified semantic abstraction layer. The framework integrates knowledge graphs, Schema.org-based semantic annotation, and a multi-agent architecture to achieve over 90% mapping accuracy across diverse domains. Compared with conventional ETL pipelines or manual mapping approaches, our method significantly improves efficiency and scalability in cross-system data integration. It provides a practical, scalable technical pathway for large-scale enterprise semantic interoperability, enabling robust, ontology-driven data unification without extensive human curation.

Large language models (LLMs) frequently exhibit hallucination errors in Text-to-SQL tasks—such as generating invalid column names or mismatching values to incorrect columns—due to insufficient domain knowledge of database schemas (e.g., table/column names) and real-world cell values. To address this, we propose a database knowledge injection framework that jointly models structured schema information and authentic cell values as prior knowledge, pre-training LLMs to internalize this relational semantics. This is further enhanced via downstream fine-tuning and schema-aware prompting. Our approach significantly improves semantic understanding and generalization across diverse database schemas, achieving state-of-the-art exact match (EM) and execution accuracy (EX) on multiple benchmarks—including Spider, Bird, and DuSQL. It effectively mitigates column name hallucination and value-column alignment errors, while demonstrating strong cross-database transferability without requiring task-specific retraining.

Automatically aligning unstructured text—such as clinical case notes—with diverse, heterogeneous database schemas remains challenging due to rigid schema-binding assumptions in existing approaches. Method: We propose a model-agnostic text structuring method grounded in an attribute grammar metamodel (MetaG), which constructs semantically enriched parse trees. Through iterative tree rewriting and extraction, the method jointly generates executable semantic schemas and their corresponding instances—without requiring prior knowledge of target data models. Contribution/Results: Our key innovation lies in decoupling text understanding from database schema design, enabling seamless adaptation to heterogeneous storage backends—including relational, graph, and document databases. Empirical evaluation on clinical texts demonstrates accurate generation of cross-model-compatible schemas and instances, significantly overcoming limitations of conventional schema-bound paradigms. The approach achieves robust schema generalization and instance fidelity across disparate target models, establishing a foundation for flexible, database-agnostic text-to-structure translation.

Weak semantic validation in Text-to-SQL systems—where existing methods prioritize syntactic correctness over alignment between user intent and SQL semantics—limits reliability and interpretability. To address this, we propose HEROSQL, a novel framework featuring: (1) a dual-granularity hierarchical representation that jointly encodes global logical plans (LPs) and local abstract syntax trees (ASTs); (2) a nested message-passing neural network (NMPNN) for cross-level semantic aggregation between LPs and ASTs; and (3) an AST-driven sub-SQL augmentation strategy to generate high-quality negative samples. Evaluated on both in-domain and out-of-domain benchmarks, HEROSQL achieves +9.40% AUPRC and +12.35% AUROC over state-of-the-art methods. Moreover, it enables fine-grained localization of semantic errors, thereby enhancing the granularity of large language model feedback and improving query platform interpretability.

Existing Text-to-SQL systems struggle with semantic ambiguity and limited scalability in complex enterprise databases due to their reliance on static schema representations. This work proposes APEX-SQL, a novel framework that shifts the Text-to-SQL paradigm from passive translation to active exploration. During schema linking, APEX-SQL integrates logical planning, dual-path pruning, and parallel data profiling to generate hypotheses, which are then validated through global topological synthesis. In the SQL generation phase, it employs a deterministic mechanism to retrieve exploration instructions, thereby enhancing semantic accuracy. The approach significantly improves reasoning capabilities over complex databases, achieving execution accuracies of 70.65% on BIRD and 51.01% on Spider 2.0-Snow—outperforming current baselines while reducing token consumption.

Existing natural language to SQL (NL2SQL) approaches often struggle to generate semantically correct and executable queries in multi-dialect database environments due to dialect coupling, semantic degradation, or cross-dialect interference. This work proposes Dial, a novel framework that decouples intent understanding from dialect-specific syntax by introducing dialect-aware logical query planning, a hierarchical intent knowledge base (HINT-KB), and an execution-driven semantic validation loop. These components collectively enable precise translation from natural language to dialect-specific SQL. Evaluated on the newly constructed DS-NL2SQL benchmark, Dial outperforms state-of-the-art methods by 10.25% in translation accuracy and achieves a 15.77% improvement in dialect feature coverage.

This work addresses the challenges of natural language to SQL (NL2SQL) translation in real-world enterprise databases, where complex table schemas, opaque column names, dialect heterogeneity, and deeply nested queries hinder performance. To tackle these issues, the authors propose a semantic-layer mediation mechanism that introduces Semantic Model Queries (SMQ) as an intermediate representation, decoupling user intent from physical SQL generation. They further design a constrained think-execute loop and a deterministic compiler to prevent overfitting to the raw database schema. Built upon the Gemini 3 Pro large language model and supporting SQLite, BigQuery, and Snowflake backends, the system achieves a 94.15% execution accuracy on the 547 tasks of Spider2-snow, ranking third on the official leaderboard and significantly outperforming approaches that rely solely on the original schema, thereby substantially enhancing cross-dialect NL2SQL generalization.