Bayesian Model Fusion for Forecasting Civil Unrest

Abstract

With the rapid rise in social media, alternative news sources, and blogs, ordinary citizens have become information producers as much as information consumers. Highly charged prose, images, and videos spread virally, and stoke the embers of social unrest by alerting fellow citizens to relevant happenings and spurring them into action. We are interested in using Big Data approaches to generate forecasts of civil unrest from open source indicators. The heterogeneous nature of data coupled with the rich and diverse origins of civil unrest call for a multi-model approach to such forecasting. We present a modular approach wherein a collection of models use overlapping sources of data to independently forecast protests. Fusion of alerts into one single alert stream becomes a key system informatics problem and we present a statistical framework to accomplish such fusion. Given an alert from one of the numerous models, the decision space for fusion has two possibilities: i) release the alert or ii) suppress the alert. Using a Bayesian decision theoretic framework, we present a fusion approach for releasing or suppressing alerts. The resulting system enables real-time decisions and more importantly tuning of precision and recall.

Event Prediction in Big Data Era: A Systematic Survey

“…Then, due to the diversity and complementary nature of the different models, the fusion of the predictions from different models will eventually result in a high recall. A Bayesian fusion-based strategy has also been investigated [95]. Another system named Carbon [108] also leverages a similar strategy.…”

Correlated model fusion

“…Pseudocode for the implementation of our algorithm is presented next. A simplified version of this algorithm, excluding sampling the GGM, is presented in . Steps 1–3 draw heavily on the procedure detailed in .…”

“…Here, we will explain the basics of each model. Additional details some of the models are provided in , where the models are used in a different framework. The historical protest model is a backward looking model that issue predictions corresponding to event–location pairs that occur with high frequency.…”

Pachinko Prediction: A Bayesian method for event prediction from social media data

“…At that time, sophisticated learning approaches employing the fusion of multiple model outputs into a single prediction were found to be particularly effective, with the EMBERS (Early Model Based Event Recognition using Surrogates) model (Ramakrishnan et al, 2014) ultimately providing the best predictions according to the evaluation metrics set for the contest. A key distinguishing feature of this model compared with other machine learning approaches was the use of multiple models combined via fusion (Hoegh et al, 2015), along with a novel method for suppressing spurious model outputs.…”