A Fast Document Classification Based on the k-Nearest Neighbor
Zeng Hai · Mini-micro Systems · 2004
According to the test document distributing in the sample classes, We discover two important properties of the kNN, which can decease greatly the amount of computation during the classifying and give two algorithm EXCLUDING CLASS and INCLUDING CLASS on the based of the two properties. In addition, a fast web document classification paradigm has been presented. Our experimental results illustrate this paradigm can increase the classification speed dramatically while not decreasing its precision of original kNN classification.