A.L. Butov, A.E. Zyornushkin, А.В. Sokolov, I.V. Chebotar, A.A. Nyrtsov

The data sources for modern information systems are databases and files of various formats that usually have a unique logic structure. All that brings about a problem of storing the information of various structures into the storehouses of data since it is necessary to adapt the structure of the data being stored to the structure of the storehouses of data. For the same reason the search in such a storehouse of data becomes difficult: attributes that have similar meanings are stored separately but during the search only distinctly given attributes are taken in to account. There are several types of storehouses of data, such as catalysts of production and system of aggregation of the Internet content in which the organization of indistinct (approximate) search is expedient. To eliminate the above-mentioned drawbacks it is recommended set to use the predicable-aligned approach for designing storehouses of data. It implies the construction of information systems on the basis of four-level architecture: a data storage level, an indistinct conclusion level, a level of preliminary processing of inquiries and an applied level. The specific feature of this architecture is that it has an anthology of attributes realized in a semantic net with the description of how close the concepts are and of the "part-whole" type relations. The use of the anthology of attributes makes it possible to increase the volume of the search while slightly reducing the accuracy of the results received. A storehouse of data built according to this architecture will make it possible to solve a problem of adding additional data and attributes of objects without changing the structure of the storehouse while allowing to increase the effectiveness of work of terminal users when they solve the problems in which the volume of the search is more important than the accuracy of the results received.