Sentiment Analysis and Argumentation Mining in UNSC Speeches

This project entails sentiment analysis and argumentation mining into the recently published UN security council speeches (UNSC) corpus. The UNSC corpus contains ~65,000 UN security council speeches from ~5,000 security council meetings from years 1995-2017. Each meeting is split up into the various speeches given by member countries. Furthermore, speeches are annotated with dates, topics and overall meeting outcomes. The UNSC corpus is, however, not annotated for argumentation structures and sentiment polarities. In this project, we attempt to produce automatic machine-driven sentiment and argumentation annotations for the UNSC corpus; which could aid future human-driven annotations.