Information Entropy Based the Stability Measure of User Behaviour Network in Microblog

The spread and evolution of web hot events is due to the interaction between users. And the behaviour of users is particularly important in micro log. The information spread through the behaviour of users and information exchange among users lead to event evolution. So the behaviour of users plays an important role in information spread and evolution of web hot event. The analysis of behaviour is a key method to obtain the development direction and trend of web hot events. However, it is a challenge to quantify the behaviour of users, especially to properly predict the direction and trend of web hot events according to the behaviour of users. This paper proposes a way to build information spread network according to behaviour of users and have a statistics of degrees of node in network. Then we use information entropy to measure the stability of behaviour network according to node degrees. And we can predict the direction and trend of web hot events by their information entropy. Experimental results show that information entropy can efficiently measure the stability of network, popular events have high entropy and their structure of network is unstable, normal events have low entropy and their structure is stable, which means events with high entropy are more likely to be popular events and may evolution in different directions.