Curating identifiable data for sharing: The databrary project

Abstract

Video captures the nuances and dynamics of human behavior more cheaply and completely than other recording methods. Although the detailed fidelity of video provides significant potential for reuse, this potential goes largely unrealized because videos are rarely shared. Moreover, the large size of video files, diversity of file formats, and incompatible software tools pose technical challenges, and recordings of faces and voices pose issues regarding participant identifiability. The Databrary (databrary.org) data library, based at NYU, has developed solutions for: securely sharing research video while respecting participants' privacy, storing and streaming shared video, and managing video datasets and associated metadata. Video data are big data, and interest in recording, analyzing, and sharing video for research, education, and policy purposes continues to grow. Databrary makes video data sharing convenient and attractive for researchers, thereby increasing transparency and enhancing the potential for discovery.