摘要
设计了一种基于VSM模型的动态文本分类器,它能针对文本的不同类别建立不同的特征子空间,各特征子空间之间相互独立,同时能将文本分类中常用的2个评估指标召回率和精确率转化为正确分类率和错分率;考察了特征子空间的维数和判定界值对这2个指标的影响.该动态文本分类器能对用户输入的文本流进行动态分类.
In this paper a dynamic text classifier based on VSM is designed,it provides different classes with different feature sub- spaces. Every feature sub- space is independent of another, and can transform the two comnonly- used performance measures- recalling rate and precision rate to correct classification rate and incorrect classification rate at the same time. Also it can review the dimension of the feature sub- space and boundary value influence on these performance measures. The dynamic text classifier can categorize the user input stream of Chinese text dynamically.
出处
《河海大学常州分校学报》
2004年第2期39-43,共5页
Journal of Hohai University Changzhou
关键词
VSM模型
特征子空间
动态文本分类
特征词
VSM model
feature sub- space
dynamic text categorization
feature words
