CIF21 DIBBs: EI: Data Laboratory for Materials Engineering

该项目直接解决了材料基因组计划的目标--加快发现和部署先进材料系统的步伐。为了获得发现新材料和研究现有材料的真知灼见，科学家和工程师严重依赖于数量不断增加的材料研究数据库和几十年前的学术研究出版物。新材料的创新往往需要数年甚至数十年的时间才能开发出一种新材料。该项目通过几个步骤应对挑战，包括自动从相关电子出版物提取数据，以支持比较和分析的格式存储数据，开发先进的计算工具以改进分析，将这些工具整合到数据实验室中，以支持发现材料性能之间的新趋势和关系，并预测具有所需性能的新材料。在该项目下开发的基础设施构件使研究人员能够(1)使用文件处理技术处理科学出版物和材料科学科学数据库中的数据，以创建知识库；(2)利用机器学习技术学习这一增强知识库中的数据，以处理材料科学和工程领域的各种使用案例；(3)利用创新的信息检索和可视化工具进行深入分析，促进更快地发现新材料。这些工具将通过建立在HubZero平台上的网络门户托管和传播，该平台还将向用户提供查询和可视化数据以及进行模拟和实验的能力。数据实验室门户还提供了使用构建块在高性能计算集群上运行模拟和实验的能力。数据实验室为材料信息学提供了一个平台，能够预测金属合金的性能，并与材料发现和工程用户社区进行互动。虽然主要目标是材料研究的跨学科领域，但这些工具被设计为与领域无关，因为核心技术可以应用于广泛学科的文档和数据库，以加快科学发现的步伐。该奖项由高级网络基础设施部获得，由美国国家科学基金会数学与物理科学局(材料研究部)联合支持。

This project directly addresses the goals of the Materials Genome Initiative -- to accelerate the pace of discovery and deployment of advanced material systems. To obtain insights for the discovery of new materials and to study existing materials, scientists and engineers rely heavily on an ever-growing number of materials research databases and scholarly research publications that date back many decades. New materials innovation often takes years, sometimes decades, to develop a new material. The project addresses the challenges through several steps, including automatic extraction of data from relevant electronic publications, storage of the data in formats that support comparison and analysis, development of advanced computational tools to improve the analysis, integration of the tools into a data laboratory to support the discovery of new trends and relationships among materials properties, and to predict new materials with desired properties.The infrastructure building blocks developed under this project enable researchers to (i) use document processing technologies to process scientific publications and data from scientific databases in materials science to create a knowledge base; (ii) use machine learning technologies to learn from the data in this enhanced knowledge base to address a variety of use cases in materials science and engineering; and (iii) use innovative information retrieval and visualization tools for insightful analysis, facilitating faster discovery of new materials. The tools will be hosted and disseminated through a web portal built on the HubZero platform, which will also provide users with the ability to query and visualize data, and run simulations and experiments. The data laboratory portal also provides the ability to run simulations and experiments on high-performance computing clusters using the building blocks. The data laboratory provides a platform for materials informatics, enabling prediction of properties of metal alloys and interaction with the materials discovery and engineering user community. While the primary target is the interdisciplinary field of materials research, the tools are designed to be domain agnostic as the core technologies can be applied to documents and databases across a broad swath of disciplines to enhance the pace of scientific discoveries.This award by the Advanced Cyberinfrastructure Division is jointly supported by the NSF Directorate for Mathematical & Physical Sciences (Division of Materials Research).