Originally funded with the help of the Sloan Foundation in 1992, the Sloan Digital Sky Survey was the first major telescopic survey to publish its data under open principles. Every single image ever collected by the Survey's 2.5 meter optical telescope is available for download by astronomers, astrophysicists and other researchers. The sheer size of the data collected, however, presented its own problems. The Sloan Digital Sky Survey corpus was simply too large for every researcher to download a full copy. In response, Johns Hopkins astronomer Alex Szalay and others developed a data infrastructure that allowed astronomers to selectively query the SDSS database, extracting only those slices that were of interest to them, and which logged every database query for later documentation. To increase the usefulness of SDSS data, Szalay also built a system that allowed astronomers to upload their own datasets which could then be easily linked with the SDSS data "in the cloud" for individual analyses and for sharing with small groups of colleagues or the broader public. Funds from this grant supports efforts by Szalay to improve and expand the SDSS data infrastructure in a number of key dimensions, revamping the data uploading process to make it more user-friendly, enabling the server to extrapolate meta-data as a way to reduce time-intensive data entry, and customizing the database in ways that would make it friendlier to researchers working in other data-intensive fields, like genomics or climatology.

A "data scientist" is someone who combines mathematical and statistical sophistication, the computational skills necessary to perform hands?on analysis of data at large scale, and the communication abilities to convey results meaningfully through visualizations and narrative. This combination of skills is still quite rare and highly in demand: an oft?cited McKinsey report published in 2011 estimates that the United States will need an additional 140,000 to 190,000 data scientists by 2018. Nonprofit organizations lag industry in the use of data scientists. Even when an organization sees how data science could improve their understanding of their clients or improve the efficacy of their work, they often lack the internal expertise and resources to take action. DataKind is a new organization founded to bridge exactly this gap. Inspired by the Teach for America model, Datakind aims to connect mission?driven organizations with designers, programmers, and statisticians from the burgeoning data science community who are looking for personally fulfilling opportunities to volunteer their time and expertise. Funds from this grant will provide two-years of pilot support to DataKind to provide a consistent revenue stream while it builds a client base of organizations and volunteers, develops a sustainable business model, and cultivates long-term funding sources.

Though few deny that administrative and other large-linked datasets represent new frontiers for social science research, there have been surprisingly few formal studies that survey and document how so-called "big data" in all its forms is actually changing social science research. This grant supports a project by a team led by Eric T. Meyer at Oxford's Internet Institute (OII) to empirically document the ways social scientists are getting access to data at scale and the tools they use to work with it. Meyer and his team will conduct a series of in-depth interviews with 125 researchers and technologists in academia, industry, and government to look at a series of interrelated questions about how big data is changing research, including how data flows between data sources and scientists, what questions big data is being used to address, how does the openness of a dataset affect its use, and how public and private data are used differently by researchers.

Alongside data scientists trained in the statistical and computational methods that are integral to analysis of data at scale, there is a growing need for "digital curators," professional staff who can steward data, code, and other research products from the lab into more durable archives. Such digital curation is often discussed as one function of the future academic research library. Unfortunately, while many university libraries want to explore this new function, institutional inertia, tight budgets, and existing organizational structures have inhibited rapid change. Funds from this grant support a project by the Council on Library and Information Resources to expand its existing post-doctoral program to prepare recent science and social science Ph.D.s for positions in data curation. Grant monies will provide partial support for the development and training of a cohort of six postdoctoral students over two years. Supported students will be placed at university libraries where they will contribute to efforts to expand the institution's digital curation capabilities. Training activities funded under this grant include an initial "boot camp" that exposes participants to the current best practices in data curation, monthly professional development webinars, and an annual retreat.

This grant provides partial support for data collection, production, and distribution of a study by the National Research Council on the training of professionals in data curation. Convening a high caliber group of scientists, technologists, educators, and university administrators, the Academy will study a handful of pioneering programs around the country that have developed curricula for training students in data curation and synthesize these curricula into a set of best practices with an eye toward preparing students for the specific data curation needs of researchers in the natural and social sciences. In addition, the report will focus on quantifying future need for data curation as a profession. The report promises to provide a blueprint for other U.S. colleges and universities who wish to begin their own programs to meet the growing need for qualified, well-trained professionals with expertise in stewarding, archiving, and maintaining data.