| Popis: |
The paper is devoted to solving the task of acquiring new knowledge and revealing new dependencies on the basis of classification and further integration of information resources for the purpose of improving the effectiveness of the information processes. The relevance is justified by the significant growth of generated, transferred and processed information in business, science and the society development. The paper considers the main problems of the effective using of information resources and information processing to determine the direction of problem solving. The authors analyzed the main aspects of integrating of information in the information systems, the present state of art in the field of classification and the application of such bioinspired algorithms as artificial bee colony, ant colony, artificial immune system, particle swarm, etc. The paper proposes to perform the preliminary classification of the information resources to improve the process of their integration. The authors used the ontological models to represent the information resources. The paper presents the abstract model of solving the task of the information resources classification on the basis of representing the integration as mapping of the ontologies. To classify the ontologies, we propose two criteria of semantic similarity between the ontologies: equivalence and hierarchy. The paper describes the problem statement and the fitness functions. To solve the classification task in accordance with two criteria, we developed a two-level bagging architecture of bioinspired algorithms composition. The task is solved in terms of parallel using of several algorithms simultaneously. The authors developed a bioinspired algorithm based on the firefly swarm behavior in nature to be used in the two-level bagging architecture. The paper presents the schemes and the rules of encoding the decisions for bioinspired algorithm in terms of two levels of bagging. To estimate the effectiveness of the proposed approach, we developed a software and carried out a set of experiments on the basis of different number of the object of information resources. The criteria of effectiveness is the degree of semantic similarity between the concepts of ontologies, classified asequivalent and similar. The experiments were to compare the firefly algorithm with the greedy algorithm, which works directly with the developed rules. The results have shown that the proposed algorithm can give the effective decisions with the time complexity of O(tn2). |
