Research map / Graph learning / Graph self-supervised learning
Graph-Link: Bridging the Semantic-Structural Gap in Text-to-SQL via Constrained Subgraph Induction
Jianwei Zhong, Yuxi Yang, Quanxin Liu, Ruida Xu, Ming Chi, Yijun Mo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Steiner tree backbone induction (ablation) | Graph-Link · BIRD Dev | Hit rate: 79.6 -> 90.8 | Improved |
| add Steiner tree backbone induction (ablation) | Graph-Link · BIRD Dev | Precision: 75.5 -> 70.4 | Hurt |
| add LLM-critic pruning (ablation) | Graph-Link · BIRD Dev | Precision: 57.3 -> 70.4 | Improved |
| add LLM-critic pruning (ablation) | Graph-Link · BIRD Dev | Hit rate: 90.3 -> 90.8 | No clear effect |
| add hierarchical community indexing (ablation) | Graph-Link · BIRD Dev | Hit rate: 86.5 -> 90.8 | Improved |
| add dual-perception anchor localization (ablation) | Graph-Link · BIRD Dev | Hit rate: 87.2 -> 90.8 | Improved |
| use Graph-Link linking instead of MCS-SQL linker | fixed SQL generator · BIRD Dev | execution accuracy: 66.0% -> 68.5% | Improved |
| use Graph-Link linking instead of MCS-SQL linker on challenge queries | fixed SQL generator · BIRD Dev challenge | execution accuracy: 44.8% -> 58.6% | Improved |
| use schema linking instead of full schema | fixed SQL generator · BIRD Dev | execution accuracy: 45.7% -> 68.5% | Improved |
| use hybrid BM25+vector linking instead of full schema | fixed SQL generator · BIRD Dev | execution accuracy: 45.7% -> 29.4% | Hurt |
| vary anchor threshold between 0.55 and 0.75 (ablation) | Graph-Link · BIRD Dev | Hit rate: change of less than ±1.5% | No clear effect |
| use Graph-Link instead of MCS-SQL on prompt size | fixed SQL generator · BIRD Dev | average tokens: ∼63,800 -> ∼20,800 | Improved |
Closest papers by meaning
- Simple is Effective: The Roles of Graphs and Large Language Models in Knowledge-Graph-Based Retrieval-Augmented Generation (ICLR 2025 · similarity 0.89)
- Structure-Guided Large Language Models for Text-to-SQL Generation (ICML 2025 · similarity 0.87)
- Reasoning of Large Language Models over Knowledge Graphs with Super-Relations (ICLR 2025 · similarity 0.87)
- Alpha-SQL: Zero-Shot Text-to-SQL using Monte Carlo Tree Search (ICML 2025 · similarity 0.87)
- Think-on-Graph 2.0: Deep and Faithful Large Language Model Reasoning with Knowledge-guided Retrieval Augmented Generation (ICLR 2025 · similarity 0.86)
- Context-aware Inductive Knowledge Graph Completion with Latent Type Constraints and Subgraph Reasoning (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
