SchemaRAG: Dynamic Large Schema Reduction for LLM-driven Structured Information Extraction

Sin Yu Bonnie Ho, Arlie Coles, Erik Larsson, Eric Marshall, Nathan Bodenstab, Paul Vozila

Keywords: retrieval-augmented generation, RAG, large language models, LLMs, structured information extraction, form‑filling extraction, large-schema extraction, schema‑space retrieval, healthcare, medical, nursing, e-commerce, production latency and cost

TL;DR: SchemaRAG dynamically prunes the output schema via schema metadata and few‑shot examples (when available) for just‑in‑time, schema‑conditioned information extraction from unstructured text.

Abstract: Extracting structured data from unstructured text using large language models (LLMs) becomes challenging when the target schemas are large and complex. In such cases, including the full schema in the prompt increases cost and latency, risks lost-in-the-middle performance degradation, and can exceed context length limits. We propose SchemaRAG, a retrieval-augmented generation (RAG) framework that dynamically prunes the output schema space for schema-conditioned information extraction tasks by leveraging schema metadata and few-shot examples (when available). We evaluate SchemaRAG on real-world healthcare and e-commerce datasets. Our results show that SchemaRAG can achieve up to an 8.8% increase in micro-F1, a 47% reduction in latency, and a 48% reduction in token costs, demonstrating its practicality for large-schema extraction.

Submission Type: Emerging

Submission Number: 253