Predictive Model for Internet Public Opinion

Abstract

The Internet is becoming a spreading platform for the public opinion. It's important to model the Internet public opinion activity as accurate as possible. Hidden Markov Model (HMM) is introduced to describe the activity of the Internet public opinion. The state of the Internet public opinion is represented as the hidden state of HMM, and its characteristic is considered as the visible states in HMM. Then the HMM is used to predict the characteristic changing of the Internet public opinion using two different algorithms under the circumstances of different amount of characteristic data gathered. Experiments on the Internet public opinion are done and the result shows that the modeling method and prediction algorithms are effective.