Voting trends within the United Nations

Understanding the way countries voted in past sessions of the United Nations General Assembly could be useful in predicting how countries would vote in future sessions. This research entailed  mining a dataset containing previous voting patterns of the different countries concerning United Nations (UN) resolutions, in order to attempt predicting how each country would vote on future UN resolutions and whether a resolution would be carried or otherwise.

The dataset consulted was a publicly available dataset that covered UN resolutions from 1946 to 2021, including their description and how each country voted. A resolution within the context of the UN General Assembly is a formal expression of the opinion or will of UN member states. Voting takes place by all the countries to establish whether the resolution would be adopted as an approved resolution. The features used from the above-mentioned dataset were the set of text describing the content of the proposed resolution. This text was then converted into vectors so that it could be understood by an algorithm that would be able to issue predictions based on input features represented as numerical vectors. Multiple models were trained to predict a country’s vote on a particular resolution.

This study was divided into three main tasks, each one with its own objective. The first was to predict how each country would vote on a resolution. The second objective was to understand whether a resolution would go through, by adding up the predictions achieved by the models trained in the first objective and establish whether the resolution was actually carried or otherwise. The third objective was to determine the number of years needed to train a model in order to achieve a good level of accuracy for the first 2 objectives. 

The accuracy scores were calculated for each trained model, to measure how close the model was to actually predicting the result.

Figure 1. Block diagram describing the objectives

Figure 2. Flags of UN member states

Student: Aidan Seychell

Supervisor : Dr Joel Azzopardi