A logical-functional approach to data bases (Q802319)

The paper describes a new approach to the semantics of data bases covering not only information aspects but also manipulation aspects of data bases. The semantics of data bases is defined in terms of universal program logics [\textit{V. N. Red'ko}, Programmirovanie 1978, No.5, 3-24 (1978; Zbl 0419.68009)] using the principles of composition programming defining derivation of new programs from some previously derived or basic initial programs using special class of compositions. Data structures are formalized using the class of the so-called named data defining connection of names with the corresponding information content. Using the notion of named function, the class of data-manipulating functions (DM- functions) is defined as the class of the denotation preserving named functions (functions, having been applied to a set of denotations, yield a subset of the former set). A class of compositions is introduced that defines synthesis of DM-functions. These compositions form an imperative logic quite directly fitting program constructs commonly used in programming languages and forming a way of defining denotational semantics alternative to the fixed point semantics. Using the formalism introduced, a generalization of relational data bases, the so-called quasirelational data bases are defined and a functional base for the retrieval operations determining a proper subclass of all quasirelational DM-functions. The class of quasirelational DM-functions covers a much wider variety of retrieval operations over data bases than Codd's relational algebra. Concerning certain properties of the logics introduced, two open problems are formulated. Suitability of the approach used is demonstrated by examples of several DM-functions defining quite ordinary queries that are not expressible in Codd's algebra.