Research on Constructing a Mass Ontology-matching Model Based on Deep Learning
Ontology-matching is an effective method for data communication and operation of multi-source heterogeneous ontologies in semantic network.It is a major challenge in the design of Ontology-matching algorithm for researchers to achieve scientific classification of mass ontology and reduce the time,space and consumption during the process of heterogeneous Ontology-matching.This paper presents autoencoder and Deep Learning technology of mass Ontology-matching algorithm design which provides important support for mass ontology semantic matching and data mining to deal with this problem.It consists of semantic embedding and model training for input ontology by using autoencoder,which can reduce the complexity of heterogeneous ontology and improve the accuracy of heterogeneous ontology classification.It is clearly showed that the algorithm has obvious advantages in time complexity and operation accuracy through experiments of multiple Ontology-matching systems,which has a high value of research and application.