Core Scientific Dataset Model: A lightweight and portable model and file format for multi-dimensional scientific data

• Published in: PloS one Vol. 15; no. 1; p. e0225953
• Main Authors: Srivastava, Deepansh J, Vosegaard, Thomas, Massiot, Dominique, Grandinetti, Philip J
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 02.01.2020; Public Library of Science (PLoS)
• Subjects: Archives & records; Chemical Sciences; Computer and Information Sciences; Coordinates; Data Analysis, Statistics and Probability; Data exchange; Database administration; Databases, Factual; Datasets; Dependent variables; Earth Sciences; Independent variables; Information management; Lightweight; Management; Material chemistry; Metadata; Methods; Models, Statistical; People and Places; Physics; Portability; Programming languages; Research and analysis methods; Science - statistics & numerical data; Science Policy; Variables; United States; Denmark; Ohio; United States > US; Aarhus Denmark; databases; metadata; biology and life sciences; data management; earth sciences; data mining; data processing; computer software; information storage and retrieval; engineering; open data; data reduction; physical sciences; software tools
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0225953

The Core Scientific Dataset (CSD) model with JavaScript Object Notation (JSON) serialization is presented as a lightweight, portable, and versatile standard for intra- and interdisciplinary scientific data exchange. This model supports datasets with a p-component dependent variable, {U0, …, Uq, …, Up-1}, discretely sampled at M unique points in a d-dimensional independent variable (X0, …, Xk, …, Xd-1) space. Moreover, this sampling is over an orthogonal grid, regular or rectilinear, where the principal coordinate axes of the grid are the independent variables. It can also hold correlated datasets assuming the different physical quantities (dependent variables) are sampled on the same orthogonal grid of independent variables. The model encapsulates the dependent variables' sampled data values and the minimum metadata needed to accurately represent this data in an appropriate coordinate system of independent variables. The CSD model can serve as a re-usable building block in the development of more sophisticated portable scientific dataset file standards.