Abstract:  Based on the needs of the network public opinion monitoring, this paper designs a model for automatic discovering the network public opinion hot issues. The system includes public opinion information acquisition, Chinese word splitter, feature selection, text segmentation and clustering analysis. By improving the Kmeans algorithm, the sensitivity of the algorithm for outlier is reduced, and the time and space complexity of the algorithm is reduced also. This paper makes use of F1 value to compare the improved Kmeans algorithm with the traditional Kmeans algorithm, which obtains satisfactory results and proves the feasibility and effectiveness of this model.



 Key words:  network public opinion, VSM, improved Kmeans algorithm, hot issues