【摘要】 针对文本特征选择中原始特征空间维数过高,提出一种基于类别相关性及遗传算法的文本特征选择方法.有效地降低了特征空间的维数,提高了分类准确率.实验验证了该方法的有效性.更多还原

【Abstract】 For high-dimensional original feather space in automatic text categorization,a new feature selection method based on category relevance and Genetic Algorithm is presented.It can effectively reduce dimensions of feature space and enhance the accuracy of classification.The experiment illustrates the validity.更多还原