Introduction

Libraries have long taken responsibility for maintaining the scholarly record by selecting, collecting, organizing, preserving and providing access to publications. As data become part of the scholarly record in their own right, libraries are confronted with a new set of responsibilities. Some research libraries are curating data, some are deferring to data centres and disciplinary repositories and some are ignoring data entirely. While improving the ability to use and reuse data is a central goal of e-research programs in the UK, USA and elsewhere (Cyberinfrastructure Vision for 21st Century Discovery, 2007; Hey and Trefethen, 2005), the best ways to accomplish this goal have not been determined (Lyon, 2007). Notions of what it means to ‘publish’ data are far less mature than notions of publishing journal articles and books (Borgman, 2007). Definitions of ‘data’ vary widely between disciplines and between individual research specialties. Librarians, scholars, funding agencies and publishers are entering a new conversation about which data will be of most future use to whom, and how to capture, preserve, curate and make those data accessible over the short and long term (Borgman, in review).

Use and reuse of data

Today's scholarship is distinguished by the extent to which its practices rely on the generation, dissemination and analysis of data. These practices are themselves distinguished both by the massive scale of data production and by the global dispersion of data resources. The rates of data generation in most fields are expected to increase even faster with new forms of instrumentation such as embedded sensor networks in the sciences, mass digitization of texts in the humanities and the digitized traces of human behaviour available to the social sciences. Digital scholarship is spawning its own new set of research questions about how to manage the ‘data deluge’, about the changing nature of scholarly practices and about economic and policy models to sustain access to research data (Borgman, 2007; Hey and Trefethen, 2003).

The scholarly value chain

Data can be reused to leverage research investments, whether by replicating or verifying findings or by asking new questions with extant data. Data are even more valuable if they can be linked to the resulting publications and to other associated objects such as field notes, grant proposals and software models.