一种基于改进型KNN算法的文本分类方法

KNN算法是比较适合于文本分类的一种分类算法，但由于其计算复杂度会随着训练集规模的增加而线性增加，从而限制了它的实际应用效果。通过改变对近邻点的搜索策略，提出了一种改进型的KNN算法。该算法在对最近邻的选择过程中，放弃传统算法中遍历所有样本的做法，而是通过逐渐逼近的思想来寻找最近邻点。实验证明，该方法在保持和传统的KNN算法几乎一样的精度性能前提下，可以明显降低算法的计算复杂度，降低时间开销，取得了较满意的结果。

A text classification method based on improved KNN algorithm

KNN algorithm is a classification algorithm suitable for text classification , but because its computa-tional complexity increases linearly with the increase in the size of the training set ,its practical application is lim-ited.In this paper, we propose an improved KNN algorithm by changing the strategy of searching the nearest neighbor points .The algorithm abandons the practice of the traditional algorithm to traverse all samples in the se-lection process nearest neighbor , and utilizes the idea of gradual approximation to find the nearest neighbor .Ex-perimental results show that the proposed algorithm ,under the premise of maintaining almost the same precision performance of traditional KNN algorithm ,can significantly reduce the computational complexity of the algorithm and running time .