首页 | 本学科首页   官方微博 | 高级检索  
     检索      

一种改进的kNN方法及其在文本分类中的应用
引用本文:孙丽华,张积东,李静梅.一种改进的kNN方法及其在文本分类中的应用[J].应用科技,2002,29(2):25-27.
作者姓名:孙丽华  张积东  李静梅
作者单位:哈尔滨工程大学,计算机科学与技术学院,黑龙江,哈尔滨,150001
摘    要:介绍了基于kNN的文本分类方法,分析了kNN方法实质,指出了该方法的不足,然后指出了一种改进方法。改进方法是基于文本属性关系和概念共现等基础上提出来的。它实质上是强化了文本中语义链属性因子的作用,修正了次要因素的噪声影响,使文本分类结果更加理想,已有的测试结果证明了这一点,尤其在测试文本与训练文本集中的某些文本直观上较相似时,结果更佳。

关 键 词:属性关联  改进kNN  文本分类
文章编号:1009-671X(2002)02-0025-03
修稿时间:2001年10月13

An Improved k-Nearest Neighbor System and Its Application to Text Classification
SUN Li_hua,ZHANG Ji_dong,LI Jing_mei.An Improved k-Nearest Neighbor System and Its Application to Text Classification[J].Applied Science and Technology,2002,29(2):25-27.
Authors:SUN Li_hua  ZHANG Ji_dong  LI Jing_mei
Abstract:This paper first reported the text classifying method based on k Nearest Neighbor(kNN) algorithm and analysed the kNN's physical meanings in the Vector Space Model(VSM) and its weakness, then put forward an improved method,which is based on text attribute association and concept co-occurring. Essentially, it emphasizes the text attribute factor of semantic chain, reduces the noises of sub-factor, and increases accuracy rating. Testing result confirmed it. The result is more accurate, especially when the testing documents are more similar to the training documents.
Keywords:k Nearest Neighbor  attribute association  improved kNN  text classification
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号