Modeling complex astrophysics data

Abstract

Scientific data are often characterized by large volumes of data, complex data relationships, and complex algorithms that operate on those data. Databases at the Space Telescope Science Institute are used to illustrate some of the differences between traditional data management schemes and the requirements of scientific data. Data from radio astronomy are modeled using conceptual enhancements to flexible image transport system (FITS) binary tables, which are compared to relational database tables. These enhancements are used to capture features of a more general notion of an observation. Finally, it is argued that accessing and manipulating each data, in order to support the entire process of data analysis, requires the construction of an abstract data model that intuitively models how a domain specialist views the data.<>