Predicting Protest Events with Hidden Markov Models

Abstract

Proactive handling of protest events which are common happenings in both democracies and authoritarian regimes requires that the risk of upcoming protest-related events are continuously assessed. Most existing approaches comparatively pay little attention to consider the event development stages. In this paper, we use auto-coded events dataset GDELT(Global Data on Events, Location, and Tone) to build a Hidden Markov Models (HMMs) based method to predict indicators associated with country instability. The method utilizes the temporal burst patterns in GDELT event streams to uncover the underlying event development mechanics and formulate the protest event prediction as a sequence classification problem based on Bayes decision. Extensive experiments with data in Thailand demonstrate the effectiveness of this method, which outperforms the logistic regression method by 27%, and the baseline by 62%.