RECIPER: A Dual-View Retrieval Pipeline for Procedure-Oriented Materials Question Answering

TL;DR

RECIPER is a dual-view retrieval pipeline that enhances procedure-oriented materials question answering by combining paragraph context and LLM-extracted procedural summaries, leading to significant retrieval improvements.

Contribution

It introduces a novel dual-view retrieval method that integrates paragraph context and procedural summaries, improving retrieval performance for materials science questions.

Findings

RECIPER improves early-rank retrieval metrics across four dense retrieval backbones.

It achieves up to 86.82% Recall@1 on the dataset.

Procedural summaries enhance downstream question answering performance.

Abstract

Retrieving procedure-oriented evidence from materials science papers is difficult because key synthesis details are often scattered across long, context-heavy documents and are not well captured by paragraph-only dense retrieval. We present RECIPER, a dual-view retrieval pipeline that indexes both paragraph-level context and compact large language model-extracted procedural summaries, then combines the two candidate streams with lightweight lexical reranking. Across four dense retrieval backbones, RECIPER consistently improves early-rank retrieval over paragraph-only dense retrieval, achieving average gains of +3.73 in Recall@1, +2.85 in nDCG@10, and +3.13 in MRR. With BGE-large-en-v1.5, it reaches 86.82%, 97.07%, and 97.85% on Recall@1, Recall@5, and Recall@10, respectively. We further observe improved downstream question answering under automatic metrics, suggesting that procedural summaries can serve as a useful complementary retrieval signal for procedure-oriented materials question answering. Code and data are available at https://github.com/ReaganWu/RECIPER.