基于多策略改进蝙蝠算法的文本特征选择

Text Feature Selection Based on Multi-strategy Improved Bat Algorithm

特征选择是文本分类过程的重要处理步骤,在其他分类预处理环节和分类算法确定的条件下,通过传统特征选择方法很难大幅度提高文本分类的准确率。针对此问题,介绍了一个基于改进蝙蝠优化的新的文本特征选择方法,即利用传统的特征选择方法对原始特征进行预选,在此基础上使用高斯局部扰动和自适应调节权重机制改进传统蝙蝠群算法,并以二进制编码形式对预选特征进行优选,分类准确率作为个体的适应度,提出了多策略改进蝙蝠算法的文本特征选择算法MS-BA,实现对文本特征选择优化模型的高效求解。结果表明,采用MS-BA进行特征优选后,其分类准确率得到有效提高。

Feature selection is an important processing step of the text classification process.It is difficult to greatly improve the accuracy of text classification by traditional feature selection methods , when other classification processing and algorithms are set.Therefore , a new text feature selection method based on improved bat optimization is introduced.It uses traditional feature selection method to pre - select the original features , based on which Gaussian local perturbation and adaptive adjustment weights are used to improve the traditional bat group algorithm.The preference and classification accuracy of pre - selected features is used as the fitness of the individual in binary coding.The multi - strategy improved bat algorithm text feature selection algorithm MS - BA is proposed to realize the efficient solution of text feature selection optimization model.The results show that the classification accuracy of MS - BA is improved.