Semi-supervised classification algorithm based on the KNN

KNN algorithm is a classification algorithm based on examples. For a test documentation, we need to calculate the similarity with each text of the training sample focus, the computation complexity is very high. According to this problem, this paper puts forward a method based on the EM-KNN semi-supervised classification algorithm. Firstly, the algorithm to cluster the training set, calculate the center of each category, then combine the center of each category and the clustering the text to form new training set. Finally train the new training set with classical KNN algorithm. Experimental results show that computational complexity can be reduced largely and the performance of the classifier can be improved by this algorithm.