PcMSP

PcMSP数据集由加州大学圣巴巴拉分校创建，专注于从多晶材料合成过程文本中提取科学行动图。该数据集包含从305篇开放获取科学文章中提取的合成句子、实体提及和句子内关系，通过两步人工标注和标注者间一致性研究确保高质量。PcMSP旨在支持可重复研究、机器自动化和材料预测，通过提供详细的合成过程信息，解决材料科学领域信息提取的挑战。

The PcMSP dataset, created by the University of California, Santa Barbara, focuses on extracting scientific action graphs from texts related to polycrystalline material synthesis processes. This dataset includes synthesis sentences, entity mentions, and intra-sentence relationships extracted from 305 open-access scientific articles. Its high quality is ensured via two-step manual annotation and an inter-annotator agreement study. Designed to support reproducible research, machine automation, and materials prediction, the PcMSP dataset addresses the challenges of information extraction in the materials science domain by providing detailed synthesis process information.