CIF21 DIBBs: EI: Creating a Digital Environment for Enabling Data-Driven Science (DEEDS)

该项目创建了一个平台，通过集成多模态数据、计算软件和用于搜索和探索的交互式工具，简化了从各种科学数据集中提取知识的过程。这项工作为共享研究数据、计算和统计算法提供了一个协作环境。在研究调查期间，环境上传和组织数据，执行和跟踪计算工作，并将输入和输出与分析结果联系起来。当调查完成后，该环境有助于发布数据、算法和调查工作流程，供公众访问。该环境还简化了对已发布数据集的发现、搜索和探索——提供了用于查看和分析数据的交互式工具。四个不同的用例（化学、电子工程、营养科学和环境科学）为平台的可用性和功能提供了一个测试平台。这些用例产生了重要的数据集，供HHS、USDA、EPA、DOE和其他政府机构使用，以支持有关政策和法规的决策。该项目建立在2014年DIBBs试点示范奖（#1443017 - DataHub）的基础上，该奖项创建了一个用于发布和发现科研数据集的数据管理平台。该团队扩展了该平台，通过以下方式支持整个研究调查过程：1)将数据连接到计算和统计建模软件；2)跟踪研究工作流程以链接数据、算法和结果；3)自动捕获元数据并按类型对数据进行分类；4)提供接口以定义复杂的分层、结构化数据并使用分析工具包对数据进行操作。使用交互式工具对发布的数据集进行探索，这些工具解释数据类型，用于高级导航、查看、搜索、分析和可视化。与化学、电气工程、营养科学和环境科学研究项目的合作产生系统需求，并保证通用的、学科中立的平台为它们的用例提供端到端的支持。该项目帮助研究人员整理他们自己的发现，也促进了数据和发现的共享，以保存、复制和扩展。

This project creates a platform that simplifies knowledge extraction from diverse scientific datasets, by integrating multimodal data, computing software, and interactive tools for search and exploration.  The effort provides a collaborative environment for sharing research data and computational and statistical algorithms.  During research investigations, the environment uploads and organizes data, executes and tracks computational jobs, and connects input and output to analysis results. When investigations are complete, the environment helps publish data, algorithms and investigative workflows for public access. The environment also simplifies the discovery, search, and exploration of published datasets - providing interactive tools for viewing and analyzing data. Four distinct use cases (chemistry, electrical engineering, nutrition science, and environmental science) provide a testbed for platform usability and functionality. The use cases produce important datasets that are intended for use by HHS, USDA, EPA, DOE, and other government agencies to support decision-making about policy and regulations.The project builds upon a 2014 DIBBs pilot demonstration award (#1443017 - DataHub) which created a data management platform for the publishing and discovery of scientific research datasets. The team extends this platform to support the full research investigation process by 1) connecting data to computational and statistical modeling software, 2) tracking research workflows to link data, algorithms, and results, 3) automatically capturing metadata and classifying data by type, and 4) providing interfaces to define complex hierarchical, structured data and operate on the data using analytical toolkits. Published datasets are explored with interactive tools that interpret data types for advanced navigation, viewing, search, analysis, and visualization. Collaborations with research projects in chemistry, electrical engineering, nutrition science, and environmental science produce system requirements and guarantee that the general, discipline-neutral platform provides end-to-end support for their use cases. The project helps researchers curate their own findings, and also facilitates the sharing of data and findings for the purposes of preservation, replication, and extension.