Stewarding Big Data

Editor's Notes

• #2: Thanks for having me here today. I’m going to do my best to give you an overview from the perspective of libraries and archives on keeping big data for scholarship and public policy.
• #3: I like the term “stewarding” to sum up all the activities involved in acquiring, preserving and making available data sets. Stewarding is essential if we as a society are going to see the full potential from big data. It’s a pretty basic proposition: somebody must devote time and effort to keeping data and to helping users access it. If this doesn’t happen, data will be hard to use, scattered and even lost. There are two basic considerations here. Collecting organizations need to concern themselves with the full life cycle of data, from initial creation, through use, to “archiving,” to long-term preservation and access, and The job is bigger than any one organization can handle; the volume and complexity of data require many organizations to work together in new ways.
• #4: I thought a good way to frame this discussion would be to summarize what a variety of organizations said in response to a recent White House request for information. This request asked for input about ensuring stewardship and encouraging broad public access to federally funded scientific research data. The White House will use the information submitted to draft revised agency policies in connection with big data. This has huge potential. The revised policies could cover requirements for data management tied to billions in funding from the National Science Foundation, the National Institutes of Health and other funding agencies.
• #5: The White House says they received 118 individual responses, all of which are made available on their website. There’s an interesting mix of respondents. Half came from discipline-specific academic research departments or professional organizations. I’d characterize them as data creators and data users. About a third of the submissions came from libraries, archives and other collecting entities. The rest came from a mix of individuals, publishers and commercial organizations. What we have here is an excellent data set that offers a broad-based snapshot of current thinking on data preservation. The response data set is seriously unstructured, as it is made up of randomly formatted textual documents, but it fairly easy to analyze.
• #6: I was pleasantly surprised at the degree of congruence among the comments. Nearly everyone enthusiastically agreed that enhanced data stewardship was critical, both to support primary scientific research and broad secondary use by the public. Most submissions explicitly called for increased resources for data stewardship. There was heavy agreement that a distributed national digital stewardship infrastructure was the right vehicle for the infusion of new funding. Apart from money, the comments also aligned in calling for a strong data preservation mandate from funding agencies. The basic idea is that receipt of funding awards should be tied to a clear expectation for long-term data management. Many of you won’t be surprised to hear that there was much less agreement on traditionally thorny topics such as intellectual property, copyright and personal privacy.
• #7: In terms of a push for increased resources, the comments clustered around three intentions. Individual funding awards should include a dedicated line item for data stewardship There is a need for models for projecting the lifetime cost of keeping data. Funding also needs to be channeled to a national infrastructure, most especially to support a distributed network of data repositories.
• #8: The focus on a national data infrastructure zeroed in on ideas for extending work that’s already underway in terms of standards, tools, and best practices. There was enthusiasm for boosting the present community of practice for data stewardship, most particularly in a way that bridges different research disciplines. This really makes sense to me. While there is excellent work going on, much of it tends to reside in specialty silos. We need to accept that, at a certain level, data stewardship has a common set of requirements that are best addressed collectively. Related to this is a pressing need for a much expanded work force of data stewards.
• #9: The need for a data preservation mandate comes down to what economists call “incentivizing.” In other words, if we want better data management, principal investigators have to be properly motivated. This motivation can come in different forms. Funding applications could call for detailed attention to data management. Evaluation of funding awards can be tied to prior demonstrated success with data stewardship And there could be provisions for data stewardship specialists to support PIs.
• #10: There was strong support for what I characterize as “respect for data,” which is linked to recognizing the broad potential for secondary use. Ideas for enabling this include adoption of a citation mechanism for data sets, such as that offered by the DateCite organization. Related to this was the proposal to give the same credit for providing useful data sets as is now given for published articles. Securing this kind of credit depends on developing a new set of metrics to track data sets and their use.
• #11: It’s no shock that consensus evaporated when it came to traditional hot-button issues. Commercial interests see control of IP as critical, while data users want relaxed IP barriers. Data creators fall between these endpoints—some want more stringent control, while others see the benefit of wider use. The issue of data privacy, most especially in connection with personal data collected under IRB rules, was strongly voiced by a number of creators, some of whom said that rules essentially barred any secondary use of certain data sets.
• #12: In terms of next steps, the ball is in the White House’s court. Two interagency working groups are mulling over the comments and will use them to draft new policies governing data stewardship. As I noted earlier, we have the potential for major improvements in how federally funded data is kept and used. But I hasten to add that the outcome is still uncertain. What is clear, however, is that there is a strong consensus among data producers, users and keepers about what should happen.
• #13: Here is a list of the websites I used in developing this presentation. Thank you.