Considering the time costing of KNN because of the frequently comparison of documents and vectors set, a rapid classification algorithm based on feature space index is proposed in this paper. The algorithm takes the feature set as the index of different documents, and label the documents or select the vectors set for classification directly according to the number of the feature words in different feature set which appearing in the documents to be labeled, so it can reduce the time of comparison between the documents and the vectors set. The experimental result shows that the algorithm can enhance the speed of the original KNN under almost the same accuracy, and the larger the feature dimensionality the bigger the range of the improvement of the speed.
展开▼