一种基于新的特征选择的海量网络文本挖掘算法研究  

Research of new feature selection based massive Web text mining algorithm

作　　者：张人上[1] 曲开社[2] 

机构地区：[1]山西财经大学信息管理学院,太原030006 [2]山西大学计算机与信息技术学院,太原030006

出　　处：《计算机应用研究》2014年第9期2632-2634,2684,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61070100);山西省自然科学基金资助项目(20120005;山西省教育厅基金资助项目(20110427)

摘　　要：为了提高文本自动分类准确率,提出一种改进的蜂群优化神经网络的选择特征的文本数据挖掘算法。该算法将文本特征选择转换成一个多目标优化问题,以特征维数最少、分类正确率最高为选择标准,采用蚁群算法找到最优特征子集,最后神经网络建立文本自动分类器,进行仿真实验测试算法性能。仿真实验结果表明,提出的方法从高维文本最优文本特征,提高了文本自动分类的正确率和识别效率,是一种有效的网络文本挖掘算法。In order to improve text classification accuracy, this paper proposed a text automatic categorization method based on ant colony algorithm and neural network. Firstly, it transformed the text feature selection into a multi-objective optimization problem, and took the feature dimension and the correct rate as a feature selection evaluation criteria, and used ant colony al- gorithm to solve the multi objective optimization problems to find the optimal feature subset, and then optimized the parameters of neural network by ant colony algorithm to establish the optimal text categorizer. The results show that the simulation experi- ment arry out on the text data, this proposed method not only can quickly find the optimal feature subset of text, but also im- proves the classification accuracy and efficiency.

关 键 词：文本自动分类 特征选择 蚁群优化算法 神经网络 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP301.6[自动化与计算机技术—控制科学与工程]