A hot spot clustering method based on improved kmeans algorithm

The emerging media network, which is represented by we-media, is in rapid development stage, and the hot spot in the society are often the most able to be discovered, shared and commented by we-media. Mining hot spot from we-media can help individuals to optimize their own investment behavior, help enterprises to adjust their production and investment strategies to meet market demand, and help government to monitor public opinions and seize the opportunity to guide the healthy development of public opinions. In this paper, we made some improvements to the basic K-Means algorithm according to the characteristics of hot spot discovery. The experimental results show that the purity and F value of the clustering result using our method improve slightly.