An Clustering-based Ontology Summarization Method with Structural and Semantic Information Integration
Ran Li, Xinbang Hu · 2021
As a common technique of constructing ontologies, ontology reuse provides technical support for tasks based on ontologies. However, with data scale's continuous expansion, ontology scale expands continuously, making ontology reuse difficult. To enable ontology engineers to understand and reuse ontologies quickly and accurately, ontology summarization is proposed, which aims to generate an abridged version of the original ontology. Existing summarization methods mainly use ontologies' structural information to extract important information, while hardly uses ontologies' semantic information. To handle this issue, this paper proposed an ontology summarization method which can takes semantic information into account. At first, the proposed method transfers ontologies' concepts into high-dimensional vectors with their semantic information, fuses these concepts' structural information to combine different perspectives' contents, and then uses these vectors to calculate their mutual distances. After that, using clustering algorithm to select important concepts and output central nodes as important concepts. Further, this method defines some related calculating concepts and relation importance measurements to select paths connecting important concepts, and thus get the summarized ontologies. Experiments on real ontologies revealed that the proposed method is of higher quality.