Ontology Matching Method Based on Gated Graph Attention Model

With the development of the Semantic Web, the number of ontologies grows exponentially and the semantic relationships between ontologies become more and more complex, understanding the true semantics of specific terms or concepts in an ontology is crucial for the matching task. At present, the main challenges facing ontology matching tasks based on representation learning methods are how to improve the embedding quality of ontology knowledge and how to integrate multiple features of ontology efficiently. Therefore, we propose an Ontology Matching Method Based on the Gated Graph Attention Model (OM-GGAT). Firstly, the semantic knowledge related to concepts in the ontology is encoded into vectors using the OWL2Vec* method, and the relevant path information from the root node to the concept is embedded to understand better the true meaning of the concept itself and the relationship between concepts. Secondly, the ontology is transformed into the corresponding graph structure according to the semantic relation. Then, when extracting the features of the ontology graph nodes, different attention weights are assigned to each adjacent node of the central concept with the help of the attention mechanism idea. Finally, gated networks are designed to further fuse semantic and structural embedding representations efficiently. To verify the effectiveness of the proposed method, comparative experiments on matching tasks were carried out on public datasets. The results show that the OM-GGAT model can effectively improve the efficiency of ontology matching.