Vol 50, No 1 (2016)

The concept of homogeneous integrity constraints in database systems is introduced. The metric space for the state of database schemes with homogeneous integrity constraints is established. It is shown that many common integrity constraints that are used in practice satisfy the condition of homogeneity of integrity constraints. Such constraints include key, referential, joint, and many-to-many integrity constraints, etc. The proposed constructs allow one to design metric spaces for the state of database schemes and enable the study of integrity constraints of databases based on the methods and concepts of “continuous mathematics.”

This paper describes the development of the functional-neural-network (FN-network) formalism, which was worked out to create a range of information systems for the intelligent computer processing of heterogeneous data from various information sources and automated decision-making support systems. It considers the limitations of this formalism in solving the tasks when the time intervals at which a solution is searched for are given in the form of indefinite variables. A method is proposed for avoiding these limitations by entering knowledge about the internal structure of data (metaknowledge) on these intervals into the context of a solution and it is shown that the use of metaknowledge not only solves the problem, but also improves the efficiency of searching for a solution by attracting additional information from a knowledge base and the context of a task.

A model of a semantic shift is described, which is based on the probabilistic concept of meanings. Employing the mathematical apparatus of quantum mechanics and defining the equivalents of such physical parameters as the mass, energy, and tunneling distance helped to introduce a measure of the probability of the shift of a term into another semantic field. The experimental testing of the model was conducted for the keyword lexics of the IAEA’s INIS abstract database.