An improved algorithm on micro-blog community detection

Micro-blogging is becoming increasingly prevalent in global world, which not only subverts the traditional means of communication, but also changes the entire media environment. Discovery of micro-blog community is of great value. However, the classical community discovery algorithms are generally based on links or interests only to recognize the traditional single community and limited to detect micro-blog communities effectively. Sometimes interaction among users is characterized by user\´s social information, but it\´s difficult to obtain in micro-blog. This work introduces the social network model based on the new social-networking characteristic called "following" which is employed in micro-blog. Based on the label propagation algorithm, we adopt the users\´ relationship, which was defined as the interaction from topic and hyperlink relations, and propose a micro-blog label propagation algorithm to detect communities. The experiment results on a real-world micro-blog dataset illustrate the reasonable and effective of our method. Experiment results over a real-world micro-blog data set illustrate the effectiveness and efficiency provided by our approach.