We have reasoned hot words for each paper by analyzing the citation network. Each paper may have several keywords. However, these keywords can not reflect all of the key points in this paper and its derivative points. Hot keywords utilize the information of neighbors in the academic network. Latent Dirichlet Allocation, for instance, has some general keywords at first. In the hot words, there are many surprising points. We can not find general topics related to LDA, but also applications and methods strong associated with LDA. These words can offer researchers a great opportunity to take a insight into this paper.