Microblog bursty topic detection based on user relationship

Microblog is becoming more and more popular in people´s daily life, through which people can read and send short text messages. Considering millions of data posted every day, detecting bursty topic in real-time will be helpful for people to grasp central information. In this paper we propose a novel bursty topic detection technique based on an improved method by calculating term weight, the calculating method we used takes user weight and the number of listeners, replies and collections into account. We first use a novel aging theory to model a term life cycle, then calculate user weight through improved PageRank algorithm to express term weight, at last adopt a unsupervised learning algorithm to detect bursty topic, meanwhile provide the results and evaluation in contrast with TF-IDF (term frequency-inverse document frequency) and UF-ITUF (user frequency-inverse thread user frequency) model to prove the validity of our proposed approach.