The paper presents a possible solution to the problems of structuring data of a large volume, as well as their integrated storage in structures that ensure the integrity, consistency of their presentation, high speed and flexibility of processing unstructured information. To solve mentioned problems, the authors propose a method for developing a multi-level ontological structure that provides a solution to interrelated problems of identifying, structuring and processing big data sets that has primarily natural-linguistic forms of representation. This multi-level model is developed based on methods of semantic analysis and relative modeling. The model is suitable for the interpretation and effective integrated processing of unstructured data obtained from distributed sources of information. The multilevel representation of the big data determines the methods and mechanisms of the unified meta-description of the data elements at the logical level, the search for patterns and classification of the characteristic space at the semantic level, and the linguistic level of the procedures for identifying, consolidating and enriching data. The modification of this method consists in applying a scalable and computationally effective genetic algorithm for searching and generating weight coefficients that correspond to different similarity measures for the set of observed features used in the dataclustering model.

No abstract

-The work discusses a stage of electronic computing equipment automated design -VLSI fragments placement. This task belongs to the NP-complex class of problems. The paper presents the statement of the VLSI placement problem. It is proposed an "evolution" -"search" strategy. A new search architecture based on the proposed strategy is constructed. The principal difference of this approach is the division of the search process into two stages and the application of different methods on each of them. At the first search stage it is used a genetic algorithm. At the second search stage an annealing simulation algorithm is proposed, which allows to improve the result found by the genetic algorithm. Based on this approach, a combined algorithm has been developed. A computational experiment was carried out on test examples (benchmarks). The quality of the obtained placement is on average 3.5% higher than the placement results obtained by using the well-known Dragon 2.23 algorithm. Series of tests and experiments were carried out and showed the promise of this approach. Time complexity of the developed algorithm is represented as O (n 3 ).

No abstract