Schema-Aware Localisation (SAL): Live Schema Grounding and Hallucination Validation for Oracle NL2SQL

A new middleware layer called Schema-Aware Localisation (SAL) achieves 62.6% execution-grounded truth (EGT) on 500 TPC-H natural language questions against Oracle Autonomous Database 23c using GPT-4o-mini, up from 2.2% without schema grounding, while reducing execution failures from 97.6% to 2.6%. SAL queries Oracle's USER_TAB_COLUMNS catalog to build a live schema map and validates SQL references via the Hallucination Index (Hidx), requiring no model retraining.

arXiv:2607.22572v1 Announce Type: new Abstract: Large language models can generate fluent SQL from natural language, but on real enterprise Oracle databases they frequently fail at execution time: columns and aliases are hallucinated and dialect-specific syntax is missed, leading to ORA-00904 invalid-identifier errors. In this setting, failures are primarily due to missing schema grounding: the model cannot know which tables and columns actually exist. This paper introduces Schema-Aware Localisation SAL , a lightweight middleware layer for Oracle NL2SQL that requires no model retraining. SAL queries Oracle's USER TAB COLUMNS catalog to build a live schema map, selects a relevant table subset for each question falling back to the full schema for multi-table queries , and injects this ground-truth context into the LLM prompt. Generated SQL is then checked by the Hallucination Index Hidx , which validates every alias.column reference against the live catalog, automatically rewrites predictable prefix errors, and otherwise triggers a structured retry with itemised corrections. We evaluate SAL on 500 TPC-H natural language questions executed against a live Oracle Autonomous Database 23c instance using GPT-4o-mini. Without any schema grounding, execution-grounded truth EGT; executes and matches the reference result set is 2.2% 12/500 . A hand-written static schema hint brings EGT to 62.0%. SAL, with no manual schema curation, achieves 62.6% EGT 96% simple, 95% medium, 40.7% complex while reducing execution failures from 97.6% to 2.6%.