计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (34): 25-28.

• 博士论坛 • 上一篇    下一篇

柔性KNN算法研究

张运良1,2,张 全1   

  1. 1.中国科学院 声学研究所,北京 100080
    2.中国科学院 研究生院,北京 100039
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-12-01 发布日期:2007-12-01
  • 通讯作者: 张运良

Study on flexible KNN algorithm

ZHANG Yun-liang1,2,ZHANG Quan1   

  1. 1.Institute of Acoustics,Chinese Academy of Sciences,Beijing 100080,China
    2.Graduate School of the Chinese Academy of Sciences,Beijing 100039,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-12-01 Published:2007-12-01
  • Contact: ZHANG Yun-liang

摘要: KNN算法是一种常用的效果较好的文本分类算法。但是标准KNN算法中,近邻的数目K对所有处理文本都是一样的,而判断类别时加权的仅仅是文本之间的相似度。基于近邻序列的排序,提出了变K算法,并且结合效果较好权重算法,形成了柔性的KNN算法,提高了分类的效果。

关键词: 计算机应用, 文本分类, KNN算法, 柔性, 效果

Abstract: KNN algorithm is a common and effective text categorization algorithm.In basic KNN algorithm,the K is fixed for different processing texts,and the weights of similarity for neighbors are equal.In this paper,a flexible KNN algorithm is developed with varying-K algorithm and weighting algorithm,which improves the effect of text categorization.

Key words: computer application, text categorization, KNN algorithm, flexible, effect