The Format Model

Abstract

A mathematical theory for the study of data representation in databases is introduced and developed. The theory focuses on three data constructs (collection, composition and classification). "Formats" with semantically rich yet tractable structure are built recursively using these constructs. Using formats, we obtain several nontrivial results concerning notions of relative information capacity and restructuring of data sets. As such, the format model provides a new approach for the formal study of the construction of "user views" and other data manipulations in databases.