TweetSpy: A Machine Learning Approach to Detecting Political Unrest Using Twitter

Abstract

The popular uprisings in a number of countries in the Middle East and North Africa in the Spring of 2011 were enabled in large part by local populations access to social networking services such as Twitter and Facebook. This thesis attempts to use language independent features of Twitter traffic mentioning different countries to distinguish between countries that are politically unstable and others that are stable. Towards this end, we collected several data sets of countries that were experiencing political unrest during the period now known as the Arab Spring, as well as a set of countries that were not. Several different methods are used to model the flow of information between Twitter users in data sets as graphs, called information cascades. Na¨ıve Bayesian, Support Vector Machines (SVM) and Bayesian logistic regression classifiers are applied to all data sets. By using the dynamic properties of information cascades, Na¨ıve Bayesian and SVM classifiers both achieve true positives rates of 100%, with false positives rates of 3% and 0%, respectively.