In this poster, we investigate the characteristics of online commenting behaviors which can influence on the popularity of various online contents, such as user created contents. To understand the behaviors, we make three dataset, which are crawled from two online discussion forums and one online community. We, first, analyze our dataset in aspects of thread lifetime and the number of comments per thread. Then we cluster and classify threads with two characteristics of the lifetime and the number of comments. Finally we study the characteristic of the number of comments per user.