K-近邻法的文本分类算法分析与改进  

Analysis and Improvement of Text Sorting Algorithm based on KNN Method

作　　者：于一[1] 

机构地区：[1]中北大学电子与计算机科学技术学院,山西太原030051

出　　处：《火力与指挥控制》2008年第4期143-145,共3页Fire Control & Command Control

摘　　要：文本自动分类的相关技术是数据挖掘的一个分枝,K-近邻法是常用的文本分类算法之一。在研究K-近邻法的基础上,根据其使用情况分析了K-近邻法的不足之处,在满足一定的判定函数前提下,提出改进后的K-近邻法。改进后的K-近邻法避免了K值的搜寻过程,降低了计算复杂性,提高计算效率。最后通过实验证明了改进K-近邻法的有效性。Text sorting is a branch of data mining technology and K-Nearest Neighbor （KNN） is one of the most popular text sorting algorithms. Based on research about the K-Nearest Neighbor algorithm, according to the disadvantage of KNN, this paper proposes an improved KNN method when certain decision functions are satisfied in accordance with disadvantages of the KNN method. The improved KNN avoids the searching process of k value, decreases computational complexity and improves computational efficiency. The effectiveness of the improved KNN is proved by the experiment

关 键 词：文本分类 K-近邻法 改进 怀卡托智能分析系统 有效性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]