Cost-Efficient RAG for Entity Matching with LLMs: A Blocking-based Exploration

TL;DR

This paper introduces CE-RAG4EM, a cost-efficient retrieval-augmented generation system for entity matching that reduces computational overhead through blocking-based batch retrieval, maintaining high accuracy.

Contribution

The paper proposes a novel blocking-based RAG architecture and a unified framework for analyzing and optimizing RAG systems specifically for entity matching tasks.

Findings

CE-RAG4EM achieves comparable or better matching quality than strong baselines.

It significantly reduces end-to-end runtime in entity matching tasks.

Key configuration parameters balance performance and computational overhead.

Abstract

Retrieval-augmented generation (RAG) enhances LLM reasoning in knowledge-intensive tasks, but existing RAG pipelines incur substantial retrieval and generation overhead when applied to large-scale entity matching. To address this limitation, we introduce CE-RAG4EM, a cost-efficient RAG architecture that reduces computation through blocking-based batch retrieval and generation. We also present a unified framework for analyzing and evaluating RAG systems for entity matching, focusing on blocking-aware optimizations and retrieval granularity. Extensive experiments suggest that CE-RAG4EM can achieve comparable or improved matching quality while substantially reducing end-to-end runtime relative to strong baselines. Our analysis further reveals that key configuration parameters introduce an inherent trade-off between performance and overhead, offering practical guidance for designing efficient and scalable RAG systems for entity matching and data integration.