EAGER: Secure Workflow Provenance for Collaborative Data Analytics

协同数据分析已成为大数据时代的必然和趋势。在这种协作环境中，知识产权保护机制对于维持和鼓励研究伙伴关系至关重要。这种机制不仅应该保护数据源和数据分析算法，而且还应该保护数据的来源，即数据处理历史。例如，参与方可能要求确保他们对各种数据产品(源、中间和最终)、处理步骤及其相互依赖关系的访问和共享。然而，现有机制没有提供对多步骤数据分析过程(工作流)来源的这种细粒度保护。为了应对这一挑战，本项目旨在研究和探索新的机制来确保协同科学工作流来源的访问和查询的安全，本项目的技术目标是深入了解面向数据流来源的访问和查询机制的可行性。这项高风险、高回报的工作将产生以下两个结果：(1)用于来源收集的多级细粒度安全来源访问和查询机制，包括敏感数据以及数据、任务和用户之间的敏感依赖关系；(2)自动化分析算法，以确保来源访问和查询策略符合对不断发展的依赖的期望约束。预期的技术和工具将在基因组数据分析领域进行评估，以证明其在协作数据分析方面的可用性和重要性。预期的技术将被配备到NSF赞助的协作科学工作流工具中，以实现安全的数据分析协作。

Collaborative data analysis has become a necessity and trend in the era of big data. In such collaborative environments, intellectual property protection mechanisms are critical to maintain and encourage research partnerships. Such mechanisms shall protect not only data sources and data analysis algorithms, but also protect data provenance, i.e., data processing history. For example, participating parties may request to secure their access and sharing of various kinds of data products (source, intermediate, and final), processing steps, and their inter-dependencies. However, existing mechanisms do not provide such fine-grained protection on multi-step data analytics procedure (workflow) provenance. To address such a challenge, this project aims to study and explore novel mechanisms to secure the access and querying over collaborative scientific workflow provenance.The technical goal of this project is to understand in depth about  the feasibility of dataflow provenance-oriented access and querying mechanisms. This high-risk, high-reward work will produce the following two outcomes: (1) a multi-level fine-grained secure provenance access and querying mechanism for provenance collection, including sensitive data as well as sensitive dependencies between data, tasks, and users, and (2) automated analysis algorithms to ensure that provenance access and querying policies would conform to desirable constraints on evolving dependencies. The intended techniques and tool will be evaluated in genomic data analysis domain to demonstrate its usability and significance in the context of collaborative data analytics. The expected techniques will be equipped to the NSF-sponsored collaborative scientific workflow tool for secure data analytics collaboration.