基于超图聚类的大本体分块研究  

Large-scale ontology block matching based on hypergraph clustering

作　　者：梁烽[1] 蔡淑琴[1] 

机构地区：[1]华中科技大学管理学院企业商务智能研究所,湖北武汉430074

出　　处：《广西大学学报（自然科学版）》2013年第6期1418-1423,共6页Journal of Guangxi University（Natural Science Edition）

基　　金：国家自然科学基金重点资助项目(70731001)

摘　　要：针对当前在大规模本体分块方面存在的不足,设计了一种基于超图聚类的改进分块方法,对本体进行预处理,将本体层次有向图进行超图正则化,把本体分块问题转换成图分割问题;用超图分割法进行聚类;根据聚类的结果进行自动分块。该算法不需要手动输入分块的个数,理论分析和实验结果表明,提出的分块方法能取得很好的结果。In the field of large-scale ontology mapping research, it is otten needea to paruution, large-scale ontology into blocks. Due to the shortcomings in current large-scale ontology partition, an improved partitioning method based on hypergraph clustering is designed in this paper. Firstly, the ontology is regularized into hypergraph structures by preprocessing. So, the ontology partition ques- tion is turned into a graph partition problem. Then hypergraph partitioning method for clustering is used. Finally, partition of the ontology based on the results of clustering is done automatically. The algorithm does not need the parameter of block number. The results of theoretical analysis and experiment show that the proposed partitioning method can obtain good results.

关 键 词：聚类 超图分割 大本体分块 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]