摘要
提出基于HowNet词汇相关性的聚类方法,该方法通过统计学的Z分数来消除孤立点,根据文档的稀疏分布程度,选择初始聚类中心,并且考虑词与词的相关性和词与词的语义相似性,使得文本聚类的精确度得到了提升,时间消耗上也大大减少.
The vocabulary correlation method based on HowNet is given out in this paper.This method removes isolated points through statistical z-score.The initial clustering centers are selected according to the sparse division of docunents.The relevance and semantic similarity of words are studied,which improve the precesion of docunent clustersing and reduce the time consumption.
出处
《微电子学与计算机》
CSCD
北大核心
2015年第4期90-93,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(61202227)
关键词
知网
词汇相关性
Z分数
义原
孤立点
聚类
HowNet
vocabulary correlation
z-score
sememes
isolated point
clustering