Authors: Juan Cuadrado, Elizabeth Martinez, Anderson Morillo, Daniel Peña, Kevin Sossa, Juan Martinez-Santos, Edwin Puertas
Venue: Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023)
DOI: 10.18653/v1/2023.semeval-1.214
Links: PDF
Abstract
Persuasive messages are increasingly prevalent on social networks, raising concerns across various communities. This study focuses on detecting news genre categories, framing, and persuasion techniques using lexical features. The proposed strategy leverages lexical weirdness to classify news articles as opinion pieces, factual reports, or satire. The approach was applied to subtasks 1 and 2 of SemEval 2023 Task 3.