基于闭合频繁Induced子树的GML文档结构聚类-南京师范大学学报.PDFVIP

9 2 () Vo.l 9 No. 2 2009 6 JOURNAL OF NAN JING NORMAL UN IVERSITY ( ENG INEERING AND TECHNOLOGY EDIT ION ) Jun, 2009 基于闭合频繁 Induced子树的GML文档结构聚类 苗建新, 吉根林, 朱颖雯 (, 210097) [ ] GMLMCF CLU. , Induced - , , GML Induced, Induced, . , , . M CF CLU , . - [ ] Induced, GML, [ ] TP 311 [] A [ ] 1672-1292( 2009) 02-0061-04 Clustering GML Docum ents by Structure Based on Closed Frequent Induced Subtrees M iao J ai nx in, J i Genlin, Z hu Y ingwen ( Sc ool of Computer Sciences, N anjing Norm alUniversity, Nanjing 210097, C ina) Abstract: T is paper presents an algorit m MCF CLU for clustering GML docum ents by structure. Different from ot er - algorit m s, it goes on clustering based on t e closed frequent induced subtrees, and doesn tneed comparing t e smi ilari- ty bewt een trees. T e closed frequent induced sub trees of all t e GML docum ents are com puted. T e representative closed frequent induced sub tree of every document is obtained. T e docum entsw ic ave t e sam e representative tree are regarded as a cluster. During t e clustering process, not on ly t e num ber of clusters can be obtained autom atically, bu t t e description of t e clusters can be ac ieved. By t e way, t e isolated points of t e docum ents can be found. T e expermi ental resu lts s ow t atMCF CLU is effective, and t at its perform ance is superior to t ose of ot er GML cluste- - ring algorit m s. K ey