Research on ontology incremental construction based on semantic graph

Xin Wang,Dan Zhang,Ying Wang
2010-01-01
Journal of Computational Information Systems
Abstract:Ontology construction is a complex and complicated project, which needs to describe the hidden knowledge in the mountains of information with explicit ontology language. Therefore, how to efficiently construct ontology, especially domain ontology, is becoming a hot research. In this paper, a new method of ontology incremental construction based on semantic graph is proposed, which contains two core models: Concept Extraction Model (CEM) and Relationship Extraction Model (REM). CEM is used to acquire Quasi-terms by Quasi-term window, calculate the similarity threshod between ontology and Quasi-terms, and add these Quasi-terms which have higher similarity threshod to candidate concept set. REM is used to annotate concept with Ontology Concept Set (OCS) and Candidate Concept Set (CCS), and then obtain binary frequent itemsets from related corpus by the improved Apriori algorithm, identify the relationship type of concepts in binary frequent itemsets and calculate the weight of binary relationship, lastly, construct semantic relationship graph and generate the final ontology concept tree. Experiment results show that this method is feasible and effective. © 2010 Binary Information Press October, 2010.
What problem does this paper attempt to address?