Neural Network Based Attention Degree Prediction for Internet Incidents in One-Crest Period

摘要

Observing an Internet incident, we find that its attention degrees develop in multiple wave crests. We propose a basic model to predict the trend of one wave crest based on back propagation (BP) neural network. Simulation experiments show that our model can predict one-crest trend of an Internet incident under the assumption that its maximum attention degree can be estimated. Our work can serve as an auxiliary tool for social or commercial workers to make decisions based on public opinions.