用摘要+段落双视角检索，提升材料合成步骤问答准确率。
RECIPER: A Dual-View Retrieval Pipeline for Procedure-Oriented Materials Question Answering
- 双视图检索：同时索引原文段落和LLM提取的流程摘要
- 召回率最高提升3.73，关键指标均超基线模型
- 适合需要精准查找实验步骤的研究者使用
从材料科学论文中检索程序性证据难度大，因关键合成细节常分散在长篇、上下文复杂的文档中，且仅依赖段落级稠密检索效果不佳。我们提出RECIPER，一种双视图检索管道，同时索引段落级上下文与大语言模型提取的紧凑流程摘要，并通过轻量级词法重排序融合两路候选结果。在四个稠密检索基线模型上，RECIPER均显著提升早期排名检索性能，平均实现Recall@1提升+3.73，nDCG@10提升+2.85，MRR提升+3.13。使用BGE-large-en-v1.5时，其Recall@1、Recall@5、Recall@10分别达到86.82%、97.07%、97.85%。下游问答任务自动评估也显示性能改善，表明流程摘要可作为程序性材料问答的有效互补检索信号。代码与数据见https://github.com/ReaganWu/RECIPER。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。