Bi-path Evolution Model for Online Topic Model Based on LDA

There are a large number of time-sensitive texts as data streams to be processed in open-source intelligence analysis. We design a new bi-path evolution model based online-LDA(BPE-OLDA) for the time-limited text streams.This model takes consideration of both content and intensity influences to model the composition process of human successfully. When estimating the parameters of this model, we simplify the Gibbs sampling. Experiments show that BPE-OLDA performs better than other approaches over time-limited text streams.