Improving the k-NN and applying it to Chinese text classification

With the problems of applying k-NN to Chinese text classification, this paper gives some improvements on k-NN. Word segmentation based on dictionaries and statistics can increase the accuracy of the classification and reduce the number of dimensions. Applying genetic algorithm to learn the value of k can improve classification automatization. The gradual classification mode is good for improving classification efficiency. The experiment shows that those improvements on k-NN can improve the efficiency of Chinese text classification while maintain the higher accuracy.