Exposing propaganda: an analysis of stylistic cues comparing human annotations and machine classification
Preprint
Abstract
This paper investigates the language of propaganda and its stylistic features. It presents the PPN dataset, standing for Propagandist Pseudo-News, a multisource, multilingual, multimodal dataset composed of news articles extracted from websites identified as propaganda sources by expert agencies. A limited sample from this set was randomly mixed with papers from the regular French press, and their URL masked, to conduct an annotation-experiment by humans, using 11 distinct labels. The results show that human annotators were able to reliably discriminate between the two types of press across each of the labels. We propose different NLP techniques to identify the cues used by the annotators, and to compare them with machine classification. They include the analyzer VAGO to measure discourse vagueness and subjectivity, a TF-IDF to serve as a baseline, and four different classifiers: two RoBERTa-based models, CATS using syntax, and one XGBoost combining syntactic and semantic features.Keywords
All Related Versions
This publication has 9 references indexed in Scilit:
- A novel hybrid approach for text encoding: Cognitive Attention To Syntax model to detect online misinformationData & Knowledge Engineering, 2023
- Analysing State-Backed Propaganda Websites: a New Dataset and Linguistic Study2023
- FakeNewsNet: A Data Repository with News Content, Social Context, and Spatiotemporal Information for Studying Fake News on Social MediaBig Data, 2020
- SemEval-2020 Task 11: Detection of Propaganda Techniques in News Articles2020
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding2019
- Divisive Language and Propaganda Detection using Multi-head Attention Transformers with Deep Learning BERT-based Language Models for Binary Classification2019
- Lying and Vagueness2018
- Detecting opinion spams and fake news using text classificationSecurity and Privacy, 2017
- XGBoost2016