LIIR at SemEval-2021 task 6: Detection of Persuasion Techniques In Texts and Images using CLIP features
This work addresses the challenge of identifying persuasion techniques in memes for social media analysis, but it is incremental as it builds on existing multimodal models.
The paper tackled the problem of detecting persuasion techniques in multimodal memes by combining pretrained CLIP models with chained classifiers and data augmentation, achieving ranks of 8/16 in F1-micro and 9/16 in F1-macro on the test set.
We describe our approach for SemEval-2021 task 6 on detection of persuasion techniques in multimodal content (memes). Our system combines pretrained multimodal models (CLIP) and chained classifiers. Also, we propose to enrich the data by a data augmentation technique. Our submission achieves a rank of 8/16 in terms of F1-micro and 9/16 with F1-macro on the test set.