Ontology Mapping Constructing by means of Low Rank Distance Matrix Optimization

Ontology mapping, based on similarity calculation, aims to find the similar concepts among different ontologies. In order to mathematically represent concepts, it is common to represent all information of a concept in a fixed dimensional vector. Therefore, the similarity calculation can be converted into a distance calculation between vectors, and the smaller the distance is, the larger the similarity will be. In this paper, the low-rank matrix learning strategy is used to obtain the corresponding ontology mapping strategy. The core idea of the method is to control the upper bound of the distance of the similar vertex pairs in the sample and the lower bound of the distance of dissimilar vertex pairs. At the same time, the rank of the matrix is integrated into the optimization conditions. The effectiveness of the proposed ontology trick is illustrated by the construction of ontology mapping on three ontology data.