November 2, 2012

Researchers Schooled in Big Data Management

Ian Armas Foster

With the rise of big data, advanced data management capabilities are becoming increasingly important in research circles.

Even the National Science Foundation and the National Institutes of Health have instituted data management plans as being central to any future research proposal. However, seeing as the big data problem is relatively new, many groups are left wondering how to put together an acceptable data management plan.

To help answer those questions, the Southeastern Universities Research Association teamed up with the Association of Southeastern Research Libraries to put together a “Step-By-Step Guide to Data Management.” The guide identifies six steps to creating a viable data management plan: assembling a data management toolkit, planning, collecting and checking data, describing and documenting data (generating metadata in other words), selecting a data repository, and storing and preserving data.

The guide consistently points at organizations like Databib, DataONE, and DMPTool (Data Management Plan Tool). According to the guide, DataONE is an environmental science organization committed to using big data to help the planet. Their “best practices primer” focuses on a “data life cycle” which pretty much involves the same steps the guide is trumpeting. Indeed, the guide itself notes that it was based on DataONE’s best practices.

DMPTool is a resource that allows research groups to compare data management plans of past accepted proposals and use them as a model.

One resource not to be overlooked, according to the guide, is the university library. While this may not come as a shock from a group that is half comprised of Research Library professionals, towering or expansive university libraries often have a significant amount of data to handle. Creating a database that can be searched hundreds of different ways of the myriad titles that exist is no small data feat. Further, according to the guide, libraries frequently have data management plan templates of their own.

With datasets growing larger, there is a renewed emphasis in the scientific community on storing and preserving one’s data. After all, every scientific discovery and paper has to be backed up by the evidence that drove the discovery. With as large as datasets can be, it would be easy to publish one’s results while neglecting the data and metadata involved. It would also be significantly cheaper. However, credited scientists do not think like this.

As such, it is important for the advancement of science in the big data era to have a standardized resource for data management. The “Step-By-Step Guide to Data Management” hopes to be that resource.

Related Articles

Yale Computer Scientists to Explore Big Data Developments

Researchers Target Storage, MapReduce Interactions

Research Aims to Automate the Impossible