Multi-objective particle swarm optimization for ontology alignment

In computer science design and implementation of high-tech areas in the modern society is accompanied by increasing the role of ontological knowledge base. Accumulation of shared ontologies is seen as a mechanism of unlimited knowledge acquisition about the world. However, the problem of integration, matching and alignment of ontologies is not solved yet. The problem of ontology alignment is to find such a structure and permissible parameters that provide the optimal values for one or more quality criteria. It should be noted that today there are many methods to compute the similarity between two discrete elements of different ontologies. Integration of up-to-date similarity computation techniques allows obtaining a versatile and accurate result. One of approach is based on the weights. Typically, the weights are assigned manually or by specific approaches. The main shortcoming of existing approaches is the lack of optimality. This article proposes a new combined approach for ontology alignment based on Latent Semantic Indexing and multi-objective particle swarm optimization method. For objective functions two criteria were chosen: the accuracy and recall. To obtain an optimal population the method of local search was employed to replace the worst of the population in the new generation. Experimental research of the suggested approach confirms the effectiveness of it.