Enhanced Ontology Matching for Big Data Integration

Abstract

ontology matching (OM) is a critical process for many disciplines. It aims at identifying the semantic correspondences among different ontologies that are merged for data integration. Unfortunately, OM still faces challenges, especially, in the big data integration (BDI) area. The high degree of semantic heterogeneity problem prevents the integration of relevant data and increased with large-scale ontologies of BDI. The quality of OM still needs more improvements to cope with BDI applications. So, this paper proposes a semantic OM approach called semantic matcher. It achieves the goals of semantic heterogeneity resolving and the quality improvement. It exploits the semantic similarity based on the word embedding model. The word embedding model provides efficient distribution semantic representation of domain words as vectors based on their context. The applicability of the proposed semantic matcher is evaluated through an experiment. The conference and anatomy gold standard datasets are evaluating the experimental results. Accuracy evaluates the quality through the precision, recall, and F-measure measures. Based on the experimental results evaluation, the proposed semantic matcher is promising and efficient in the semantic OM for BDI.