Mining Hot-Personae Approach Based on Local Social Microblog Graph

Abstract

With the increasing popularity of online social media platforms, netizens always chat with their friends and share information, such as what they like in their daily lives, on these platforms. Netizens publish tons of information on social platforms every day. These platforms converge many people and information. The processes by which the publishers find the sharers who are interested in their publications and the sharers find some interesting things and information in what the publishers published have resulted in the challenge of retrieving information from social network fields. To address these issues, we propose a novel algorithm, named Hot Persona Mining, to analyze the users' focus personae from microblog posts in the online social networks. During mining, we first utilize local-based graph clustering to establish the nearest neighbor nodes of target users. Then, we mine users' focused personae entities from their neighbors' published microblog posts in different periods. Then, we construct the users' active score vector and their interest matrix to mine the hot personae in every local social graph. The experimental results show that our algorithm effectively mines current focus of the target user, and exhibits good performance as shown by its precision, recall and F-measures.