The research lead by Maram Hasanain titled “News Genre, Framing, and Persuasion Techniques Detection using Multilingual Models” is scheduled to be presented at the17th International Workshop on Semantic Evaluation (SemEval) on July 13th. Hasanain’s work marks a significant step forward in the understanding and application of multilingual models, providing a lens through which we can better discern the nuances of news genre, framing, and persuasion techniques.
The task addressed three subtasks with six languages, in addition to three “surprise” test languages, resulting in 27 different test setups. It highlights the benefit of multilingual models for detecting news genre, framing and persuasion techniques detection in multiple languages. Hasanain’s study addresses different challenges, breaking them down into three subtasks that are assessed in six different languages. To add a further layer of complexity, the research incorporates three additional “surprise” test languages, which are unidentified until the testing phase. This approach results in a total of 27 distinct test setups, providing a robust foundation for the exploration and validation of the research’s findings.
The research illuminates the potential of multilingual models in identifying the genre of news, discerning the framing techniques used, and detecting the persuasion strategies employed, across multiple languages.