Reproduce “Cost-Efficient RAG for Entity Matching with LLMs: A Blocking-based Batch Processing Framework” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Chuangtao Ma, Zeyu Zhang, Arijit Khan, Sebastian Schelter, Paul Groth

The authors introduce CE-RAG4EM, a zero-shot entity-matching system that reuses retrieved factual context across candidate pairs grouped by blocking. It retrieves shared context from Wikidata, then makes pairwise or batched match decisions. Across nine benchmarks, its best retrieval-granularity variant achieves higher F1 than LLM-only prompting on all nine; this is a per-dataset best-of result, not one fixed configuration.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.