摘要
为解决MNB分类器在测试样本变化时分类精度误差较大的问题,采用权重标准补集分类器(WCNB)代替MNB分类器,并研究WCNB分类器对不同测试样本分类精度的变化,针对WCNB技术存在目标字符串变化所产生的权重计数问题,采用目标字符串频率转换技术,建立一种有误差补偿功能的WCNB分类器数学模型并进行了实验仿真.实验仿真结果验证了WCNB数学模型的可行性.
In order to solve the problems of classification accuracy that produced in text variety testing with MNB classifier, a WCNB classifier was presented to replace MNB classifier. The classification precision changes of the WCNB classifier were studied for different test samples. Due to the weight counting mistakes arisen from target strings changing in the WCNB classification, a target strings frequency conversion technology was adopted to develop a WCNB classifier mathematic model with error compensation function. The simulation result verifies the feasibility of the WCNB mathematic model.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2016年第4期382-386,共5页
Transactions of Beijing Institute of Technology
基金
国家自然科学基金资助项目(61272125)
关键词
文本分类
数学模型
计算机仿真
text classification
mathematical model
computer simulation