Abstract: The hyper accounts in micro-blog have a significant impact on the spread and development of the events in the Internet public opinion. The hidden behind hyper accounts are a special group which can’t be neglected. This paper proposes a hyper accounts finding method based on the community detecting algorithm. In this method, by constructing the bloggers association network and mining the features of the constructed network, the goal of automatically finding the hyper speculation accounts is achieved. By using the related micro-blogs of the Hunan Linwu Watermelon Death event as the experimental corpus, the blogger association network is constructed and the hyper accounts we detected and distinguished by using the proposed method, which proves the feasible of the presented method.

Key words: Internet public opinion, hyper account, community detection