In the paper topic evolution analysis is achieved by tracking the topic trends in different time-slices. Latent Dirichlet Allocation (LDA) model is built in time-slices. Gibbs algorithm is used to find out latent variables in LDA modle, Kullback-Leibler divergence is used to measure the similarity between topics. the modified Z-score method is used to measure the drift between topics in order to reflect topic evolution.