社交媒體圖像數據集

本研究构建的社交媒體圖像數據集，由以色列理工学院的研究团队创建，旨在支持社交媒體视觉内容的研究与分析。数据集包含超过200万张从Reddit和4chan收集的图像，涵盖了广泛的视觉内容，包括经过各种编辑和操纵的图像。创建过程中，研究团队不仅收集了原始图像，还生成了这些图像的多种操纵版本，以模拟社交媒體上常见的图像编辑情况。该数据集的应用领域广泛，主要用于研究社交媒體上的视觉内容传播，特别是在检测和分析视觉近重复内容方面，有助于理解和应对社交媒體上的视觉误导信息。

The social media image dataset developed in this study was created by a research team from the Technion – Israel Institute of Technology, with the aim of supporting research and analysis of visual content on social media platforms. The dataset contains over 2 million images collected from Reddit and 4chan, covering a broad spectrum of visual content including various edited and manipulated images. During the dataset construction process, the research team not only collected the original raw images but also generated multiple manipulated versions of these images to simulate common image editing scenarios encountered on social media. This dataset has wide-ranging applications, primarily focused on researching the propagation of visual content on social media, particularly in the detection and analysis of visually near-duplicate content, and aids in understanding and mitigating visual misinformation on social media platforms.