摘要
贝叶斯算法在文本分类时需要进行特征提取,传统特征提取算法存在特征提取不够准确,进而导致分类效率不高。为解决此问题,提出一种基于滑动窗口的特征选取方法,该方法能扩大特征的选取范围。实验表明,改进后的方法可以有效地提高文本的分类精度。
In text classification, Bayesian algorithm needs feature selection, but the traditional feature selection algorithm is not accurate enough, which affects classification precision. To order to resolve the problem, a new improved bayes algorithm based on slipping window method is proposed. The algorithm can extend the character's number to improve the efficiency. Experimental results showed that our algorithm performed more efficienthy than the traditional methods in classification precision.
出处
《重庆邮电学院学报(自然科学版)》
2006年第4期528-531,共4页
Journal of Chongqing University of Posts and Telecommunications(Natural Sciences Edition)
基金
新世纪优秀人才支持计划(NCET)
重庆市自然科学基金项目(2005BB2063)
关键词
朴素贝叶斯算法
滑动窗口
特征选取
邮件分类
naive Bayesian algorithm
slipping window
feature selection
mail classification