How institutional factors influence the creation of scientific metadata

This paper demonstrates how "frictions" arise in creating and managing metadata, and shows how the degree of metadata institutionalization can strongly influence data management needs and practices.

Abstract

Access to high volumes of digital data offer researchers in all disciplines the possibility to ask new kinds of questions using computational methods. Burgeoning digital data collections, however, challenge established data management and analysis methods. Data management is a multi-pronged institutionalized effort, spanning technology, policies, metadata, and everyday data practices. In this paper, we focus on the last two components: metadata and everyday data practices. We demonstrate how "frictions" arise in creating and managing metadata. These include standardization frictions, temporal frictions, data sharing frictions, and frictions related to the availability of human support. Through an illustration of these frictions in case studies of three large, distributed, collaborative science projects, we show how the degree of metadata institutionalization can strongly influence data management needs and practices.