Multimodal Stance Generation Dataset (StanceGen2024)
收藏资源简介:
StanceGen2024是一个专门为政治话语中的多模态立场可控文本生成而设计的新型数据集。由北京邮电大学的研究团队创建,该数据集汇集了2024年美国总统选举期间候选人的推文和用户评论,包含丰富的文本、图像以及视频等多模态信息。这些数据不仅涵盖了政治立场表达,还包括了与推文相关的视觉内容,为研究多模态政治内容如何影响立场表达提供了真实而多样的基础。StanceGen2024旨在推动多模态立场检测、政治话语分析以及情感分析等多种任务的研究。
StanceGen2024 is a novel dataset specifically designed for controllable multimodal stance text generation in political discourse. Created by a research team from Beijing University of Posts and Telecommunications, this dataset compiles tweets and user comments from candidates during the 2024 U.S. presidential election, encompassing rich multimodal information including text, images, and videos. These data not only cover political stance expressions but also include visual content associated with the tweets, providing a realistic and diverse foundation for researching how multimodal political content influences stance expression. StanceGen2024 aims to advance research on multiple tasks such as multimodal stance detection, political discourse analysis, and sentiment analysis.




