Can LLMs Normalize Databases? A Benchmark and Multi-Agent Framework for Schema Normalization
Dong-Jae Koh, Young-Kyoon Suh and colleagues (Kyungpook National University) introduce DNBENCH for database normalization from 1NF to BCNF and a multi-agent method, MARS, that improves the benchmark score by 82.0% over a single prompt.
Ask this paper
Benchmark: 3,275 samples across Single, Complex and Real World levels, scored on semantic equivalence, structural accuracy and logical validity.
Failures: Models repeatedly fail at dependency inference, schema decomposition and reconstructing inter-table constraints.
MARS: Separates evidence extraction, violation diagnosis and decomposition planning from schema generation and verification.
Result: MARS improves DNB-SCORE by 82.0% over the single-prompt baseline.
Abstract
Large Language Models (LLMs) are increasingly used to generate structured outputs, but their reliability remains unclear when those outputs must satisfy database-level constraints. We study this issue through database normalization, involving reasoning about functional dependencies, lossless join decompositions, and inter-table constraints. We introduce a Database Normalization Benchmark (DNBENCH), comprising 3,275 samples for evaluating LLM-driven database normalization from 1NF to BCNF. DNBENCH uses a three-axis protocol to measure semantic equivalence, structural accuracy, and logical validity. Across Single, Complex, and Real World levels, DNBENCH uncovers recurring failures in dependency inference, schema decomposition, and inter-table constraint reconstruction. We further propose Multi-Agent Reasoning for Schemas (MARS), which separates evidence extraction, violation diagnosis, and decomposition planning from schema generation and verification. MARS improves the DNB-SCORE by 82.0% over the single-prompt baseline. All artifacts will be released upon acceptance.