|Table of Contents|

On Classif ication of Associative Text Based on Rules Pruning ofMutual Information(PDF)

南京师范大学学报（工程技术版）[ISSN:1006-6977/CN:61-1281/TN]

Issue:
2008年04期
Page:
173-177
Research Field:
Publishing date:

Info

Title:
On Classif ication of Associative Text Based on Rules Pruning ofMutual Information
Author(s):
Shang B ingzhang;  B aiQ ingyuan
C ollege ofM ath em at ics and C om pu ter S cience, Fuzh ouU n ivers ity, Fuzhou 350002, Ch ina
Keywords:
 ; mutual in fo rm ation;  rules pruning;  assoc iative c lassifica tion
PACS:
-
DOI:
-
Abstract:
The traditiona l assoc ia tive c lassify ing algor ithm s of assoc iative texts gene ra te a huge mum be r of ru les. If the ru les w ere no t pruned, the e ffic iency o f c lassification would be influenced. H ow ever, if the form er prun ingm ethod were adopted, d ifferent degrees of accuracy o f c lassifica tion w ould appear. Therefore, an assoc iative text c lassification algo rithm-based on ru les prun ing o fmutual inform ation is presen ted to prune the ru les o f each c lass. The ru les w ith h igh c las s ify ing capacity are chosen to form classifiers to c lassify the texts be ing classified. The study illum inates that the mutual inform ation-based rules pruning a lgo rithm no t on ly gets much less rules but ism o re he lpfu l fo r im prov ing the accuracy o f the assoc iation categor ization. The exper imenta l resu lts show the performance o f th is m e thod is better than both ARCBC a lgo rithm and the algor ithm wh ich uses a ll rules.

References:

[ 1] Liu B, H suW, M a YM. Integ rating c lassification and assoc iation ru lem ining[ C] / / ACM In t’ l Conf on Know ledge D iscovery and DataM in ing. New Yo rk: ACM Press, 1998: 80-86.
[ 2] L iW, H an J, Pei J. CMAR: Accurate and e fficient c lassifica tion based on mu ltip le c lassification ru les[ C ] / / CerconeN. Proc
o f the 2001 IEEE Int’lConf on DataM in ing. Ca lifo rn ia: IEEE Press, 2001: 369-376.
[ 3] ZaÏane O R, An ton ieM L. C lassify ing tex t docum ents by assoc iating term s w ith tex t categor ies[ C] / / Zhou X F. Proc o f the 13th Austra lasian Da tabase Con.f M elbourne: Austra lian Com pute r Society, 2002: 215-222.
[ 4] Ag rawa l R, Sr ikant R. Fast algor ithm s fo rm in ing association ru les[ C] / / Bocca J B, JarkeM, Zan io lo C. Proc o f the 20th Vary Larg e Data Bases Conference. Santiago, 1994: 487-499.
[ 5] H an J, Pe i J, Y in Y W. M in ing frequen t pa tterns w ithout candidate generation[ J]. DataM in ing and Know ledge D iscovery,2004, 8( 1): 53-87.
[ 6] 陈晓云, 陈袆, 王雷, 等. 基于分类规则树的频繁模式文本分类[ J]. 软件学报, 2006, 17( 5): 1 017-1 025.
Chen Xiaoyun, Chen H u,i W ang Le,i et a .l Frequent pattern tex t classification based on rules tree[ J] . SofWt are, 2006, 17( 5): 1 017-1 025. ( in Ch inese)
[ 7] http: / / sewm. pku. edu. cn /QA / re ference / ICTCLAS /FreeICTCLAS /[ OL]. 中文自然语言处理开放平台网站, 2006.
http: / / sewm. pku. edu. cn /QA / reference / ICTCLAS /FreeICTCLAS /[ OL]. The S ite o f Ch inese N atura l LanÏgnage Processing Platfo rm, 2006. ( in Ch inese)

Memo

Memo:
-
Last Update: 2013-07-22