Building the Research Data Ecosystem of the Future

The extreme scaling of data is taking scientific and scholarly research to a tipping point. Rather than the traditional methodologies of personal capture, curation, and publication of data, much of the data today is captured by and for computers. A key driving factor is the rise of AI/ML as a central component of scientific and scholarly discovery: isolated data sets have less utility than joined, integrated data views that allow for training and validation. The use and reuse of data in AI/ML raises a host of challenges for data that suggest an ecosystem approach to the research data lifecycle, where an ecosystem encompasses the context in which data exists. Building a team that can drive innovation in the research data ecosystem of the future will position the IU/ORNL partnership at the center of this revolution.

The intersectional topics represented by our respective institutions include:

• Generative structured scientific/scholarly metadata
• Automating flow of research data
• Data architecture and modeling
• FAIR and automated research workflows
• Knowledge graphs, provenance
• AI/ML in research data for digital twins, clustering, labeling
• Research data engineering
• Trustable AI
• Enabling research from edge to cloud to HPC and back
• Collaboration, training, outreach, and shared workforce development
• Data lifecycle
• Research data management systems

This symposium brings together experts working in foundational research through use-inspired research to operational deployment, and with expertise in HPC, information science, data engineering, cloud computing, visualization, and education/training. The goals of the joint symposium are to share expertise and to identify complementary strengths, strategic challenges, and opportunities that have strong common interests. The anticipated outputs of this symposium are new personal and organizational connections, a report on opportunities for both institutions, and a roadmap for a new set of joint proposals and exchanges.