About This Project
For the last 11 years, the UNICEF Operation Center (OPSCEN) has been compiling a daily news bulletin pertaining to emergencies and crises throughout the world. We are interested in mining the sequential pattern between the events described in this data. For example, what is the likelihood of a coup d’état happening after a large-scale pro-democracy protest in a given country? Will it lead to other events, e.g. citizens fleeing to another country or a re-election? If yes, what is the probability?
Our approach is to analyze political and financial data to find internal correlations and patterns, and then combine the insights of each domain into an aggregate model. For financial analysis, we will use neural networks to find insightful patterns. For political data, we will tackle text mining using two types of data: Data with minimal categorization (OPSCEN’s and GDELT) and Pre-analyzed data that categorizes crisis types and grades events by level of conflict/cooperation intensity (ICEWS).
