Propagandistic Meme Dataset

该数据集包含了大约3000个来自不同社交媒体平台的梗图，这些梗图被标注为宣传性或非宣传性。每个梗图都由三名标注者进行标注，以多数决定的结果作为最终标签。数据集按照70%、10%和20%的比例分为训练集、开发集和测试集。在标签分布上，宣传性梗图占40%，非宣传性梗图占60%。该数据集的规模大约为3000个梗图，其任务是对梗图进行宣传性和非宣传性的检测，同时分析其仇恨程度。

This dataset contains approximately 3,000 meme images sourced from various social media platforms, where each meme is annotated as either promotional or non-promotional. Each meme is labeled by three annotators, with the majority voting result serving as the final label. The dataset is split into training, validation, and test sets at a ratio of 70%, 10%, and 20% respectively. Regarding the label distribution, promotional memes account for 40% of the dataset, while non-promotional memes make up 60%. The dataset has a total size of around 3,000 meme images, and its core tasks include detecting whether a meme is promotional or non-promotional, as well as analyzing the hatefulness of the memes.