FakeNVE
收藏资源简介:
FakeNVE数据集由苏州大学计算机科学与技术学院开发,包含3614条多模态新闻视频,每条视频均附有手动创建的自然语言解释。该数据集旨在通过整合视频、标题、OCR文本和相关新闻上下文等多模态信息,生成解释性句子以揭示新闻的真实性。数据集的构建过程包括从FakeSV数据集中筛选并标注新闻视频,确保每条视频的解释准确且简洁。FakeNVE数据集的应用领域主要集中在虚假新闻检测和解释生成,旨在通过多模态信息的融合,提升新闻真实性判断的准确性和透明度。
The FakeNVE dataset, developed by the School of Computer Science and Technology, Soochow University, comprises 3614 multimodal news videos, each paired with manually crafted natural language explanations. This dataset is designed to generate explanatory sentences that uncover the veracity of news by integrating multimodal information including videos, headlines, OCR text, and relevant news context. The construction of the dataset entails screening and annotating news videos sourced from the FakeSV dataset, with the goal of ensuring that each video’s explanation is accurate and concise. The primary application areas of the FakeNVE dataset are fake news detection and explanation generation, which seek to enhance the accuracy and transparency of news veracity assessment via multimodal information fusion.




