Investigating Propaganda Considering the Discursive Context of Utterances

The following system description presents our approach to the detection of propagandistic techniques in tweets. The given task has been framed as a multi-label classification problem. In a multi-label classification problem, each input chunk-in this case tweet-is assigned one of several class labels. In order to assign class labels to the given utterances, we opted for RoBERTa (A Robustly Optimized BERT Pretraining Approach) for sequence classification. Starting off with a pre-trained model for language representation, we fine-tuned this model on the given classification task with the provided annotated data in supervised training steps. In addition to the content of the message, further features describing the general communication context is taken into account.