Team QUST at SemEval-2023 Task 3: A Comprehensive Study of Monolingual and Multilingual Approaches for Detecting Online News Genre, Framing and Persuasion Techniques

To model the Task 3 (Piskorski et al., 2023) expects the participants features representation of news articles across different to develop algorithms to automatically detect languages, the XLM-RoBERTa (XLM-R) the news genre, framing and persuasion techniques (Conneau et al., 2020) is fine-tuned since it can in a multilingual setup as shown in Table 1. Six different processes all the languages existing in Task 3, and languages are covered in this task, including typically outperforms other models such as mBERT English, French, German, Italian, Polish and Russian. (Kenton and Toutanova, 2019). In addition, three surprise languages, Spanish, In addition, this study also calculates the sample Greek and Georgian, are also included in the final weights and class weights to combat the data test phase for conducting a zero-shot learning imbalance. The sample weights enable a training scenario.