NSF BCS 1238599
Administered in: Liberal Arts
Behavior develops from a variety of factors interacting across domains and time scales. To capture the richness of behavioral development and understand its complexity, most developmental scientists record video. However, researchers rarely share video data, and this has slowed progress, impeded understanding, and diminished the impact of public investments in behavioral science. The Databrary project aims to increase scientific transparency and accelerate discovery in developmental science by making it easy and convenient for scientists to share video data. No large-scale repository for sharing digital video data currently exists. So, we will create a web-based repository (databrary.org). We will develop data management tools to make contributing video files and metadata convenient and reliable for researchers, and we will create mechanisms to enable transcoding, searching, browsing, visualizing, streaming, and downloading files in the Databrary. OpenSHAPA (openshapa.org), a free, open-source video-coding tool will become the common file format for the Databrary. We will develop tools to convert data coded in other systems to the OpenSHAPA format, and we will enhance OpenSHAPA to support integration of multimedia data such as eye tracking, motion tracking, physiological measures, and neural imaging. We will address cultural, historical, technical, and privacy/ethics challenges associated with open video data sharing by building a user/contributor community of developmental scientists. With community input, we will develop template permission language so that investigators can contribute datasets to the library, and we will develop contributor and user agreements to ensure appropriate levels of access. We will provide training and technical support to the emerging user community through workshops, webinars, and on-site training.
By creating tools for open video data sharing, we expect to deepen insights and accelerate discovery in developmental science. The contribution of a particular dataset will no longer depend on the private activities of researchers from one laboratory, but instead benefit from the critique and imagination of many researchers with different viewpoints. Researchers will be able to view one another’s datasets and reanalyze them to test competing hypotheses, perform integrative analyses, learn from prior examples, and address new questions beyond the scope of the original study—enabling new possibilities for research in labs with limited financial and technical resources. Moreover, the infrastructure we develop will have broad impacts for data sharing and data management in the entire behavioral science community.