Tomorrow's Decisions Today: Using Press Releases for the Automatic Prediction of Decisions of the European Court of Human Rights
Masha Medvedeva

With more and more courts adopting the policy of publishing proceedings online, there is an ever-growing interest in automatic decision making and predicting future judicial decisions. However, with the courts publishing proceedings in full, it may be hard to identify information that may be available for analysis before the trial. The European Court of Human Rights (ECtHR) publishes press releases for some of the cases online in the week before the cases are tried, therefore giving us the unique opportunity to analyze these texts in order to predict what the decisions will be. In this paper we experiment with various Natural Language Processing techniques by analysing the wording of English and French versions of press releases published by the ECtHR. Our attempts to predict the violation of 6 articles of the European Convention on Human Rights (ECHR) show that even such short texts with minimal factual information have a potential for automatic prediction of future rulings using machine learning.