Web网页文本特征选择方法研究

Research of web text feature selection approach

提出了一种用于文本分类的特征选择方法。计算文本的特征值,每个特征值被赋予一个权重值,权重值的大小表示文本特征的重要程度,权重值最大的特征为决定性特征,决定性特征能代表某一类;通过构造树结构模型来消除噪音文本,同时还可以降低计算复杂度;最后改进该算法,动态的检测相对于当前节点的最佳节点更有利于进行特征选择。实验结果表明,该方法具有较高的分类精度,且计算成本较低,符合规模Web自动分类的需要。

A new feature selection method for text classification is proposed.Firstly,the eigenvalues of the text is calculated,the size of the weight representing the importance degree of text feature,and the greatest features among weights is the decisive feature which represent a certain category.Secondly,the noise text is eliminated by constructing a tree structure model,while reducing the computational complexity.Finally,it improves the algorithm,which dynamically detects the best node relative to the current node,and it is helpful for feature selection.Experimental results show that the method has a higher classification accuracy and lower cost on the calculation,in line with the needs of scale Web automatic classification.