“Don’t discuss”: Investigating Semantic and Argumentative Features for Supervised Propagandist Message Detection and Classification

Abstract

One of the mechanisms through which disinformation is spreading online, in particular through social media, is by employing propaganda techniques. These include specific rhetorical and psychological strategies, ranging from leveraging on emotions to exploiting logical fallacies. In this paper, our goal is to push forward research on propaganda detection based on text analysis, given the crucial role these methods may play to address this main societal issue. More precisely, we propose a supervised approach to classify textual snippets both as propaganda messages and according to the precise applied propaganda technique, as well as a detailed linguistic analysis of the features characterising propaganda information in text (e.g., semantic, sentiment and argumentation features). Extensive experiments conducted on two available propagandist resources (i.e., show that the proposed approach, leveraging different language models and the investigated linguistic features, achieves very promising results on propaganda classification, both at sentenceand at fragment-level.

Introducing a Gold Standard Corpus from Young Multilinguals for the Evaluation of Automatic UD-PoS Taggers for Italian

“…We cast this task as a sentence-span classification problem and we address it relying on a transformer architecture. Results reach SoTA systems performances on the tasks of propaganda detection and classification (for a comparison with SoTA algorithms, we refer to (Vorakitphan et al, 2021)).…”

Argument-based Detection and Classification of Fallacies in Political Debates

“…Consequently, distinct losses are computed for each model: fallacy loss (loss f al ), component loss (loss cmp ), relation loss (loss rel ), and part-of-speech loss (loss pos ). These individual losses are combined by multiplying them with an arbitrary α value of 0.1, yielding a unified average loss referred to as the joint loss (Vorakitphan et al, 2021). In our study, we opted for empirically investigating the optimal alpha value that yielded superior performance, as evidenced by our experiments (see Appendix D for the exhaustive evaluation).…”

Language Transfer for Identifying Diagnostic Paragraphs in Clinical Notes

“…To combine the results from sentence-span based RoBERTa with the feature-based BiLSTM we apply the joint loss strategy proposed in (Vorakitphan et al, 2021). Each model produces a loss per batch using CrossEntropy loss function L. Following the function: loss joint loss = α × (loss sentence +lossspan+loss semantic argumentation features ) N loss where each loss value is produced from CrossEntropy function of its classifier (e.g., loss sentence and loss span from RoBERTa models of sentence and span, loss semantic argumentation features from the BiLSTM model.…”