MASH: A Multiplatform and Multimodal Annotated Dataset for Societal Impact of Hurricane

We present a Multiplatform Annotated Dataset for Societal Impact of Hurricane (MASH) that includes 59,607 relevant social media data posts from Reddit, TikTok, and YouTube. In addition, all relevant posts are annotated on three dimensions: Humanitarian Classes, Bias Classes, and Information Integrity Classes in a multi-modal approach that considers both textual and visual content (text, images, and videos), providing a rich labeled dataset for in-depth analysis. To our best knowledge, MASH is the first large-scale, multi-platform, multimodal, and multi-dimensionally annotated hurricane dataset. We envision that MASH can contribute to the study of hurricanes' impact on society, such as disaster response, disaster severity classification, public sentiment analysis, disaster policy making, and bias identification. Usage Notice This dataset includes four annotation files: • reddit_anno_publish.csv • tiktok_anno_publish.csv • youtube_anno_publish.csv Each file contains post IDs and corresponding annotations on three dimensions: Humanitarian Classes, Bias Classes, and Information Integrity Classes. To protect user privacy, only post IDs are released. We recommend retrieving the full post content via the official APIs of each platform, in accordance with their respective terms of service. - Reddit API (https://www.reddit.com/dev/api) - TikTok API (https://developers.tiktok.com/products/research-api) - YouTube API (https://developers.google.com/youtube/v3) Humanitarian Classes Each post is annotated with seven binary humanitarian classes. For each class, the label is either: • True – the post contains this humanitarian information • False – the post does not contain this information These seven humanitarian classes include: • Casualty: The post reports people or animals who are killed, injured, or missing during the hurricane. • Evacuation: The post describes the evacuation, relocation, rescue, or displacement of individuals or animals due to the hurricane. • Damage: The post reports damage to infrastructure or public utilities caused by the hurricane. • Advice: The post provides advice, guidance, or suggestions related to hurricanes, including how to stay safe, protect property, or prepare for the disaster. • Request: Request for help, support, or resources due to the hurricane • Assistance: This includes both physical aid and emotional or psychological support provided by individuals, communities, or organizations. • Recovery: The post describes efforts or activities related to the recovery and rebuilding process after the hurricane. Note: A single post may be labeled as True for multiple humanitarian categories. Bias Classes Each post is annotated with five binary bias classes. For each class, the label is either: • True – the post contains this bias information • False – the post does not contain this information These five bias classes include: • Linguistic Bias: The post contains biased, inappropriate, or offensive language, with a focus on word choice, tone, or expression. • Political Bias: The post expresses political ideology, showing favor or disapproval toward specific political actors, parties, or policies. • Gender Bias: The post contains biased, stereotypical, or discriminatory language or viewpoints related to gender. • Hate Speech: The post contains language that expresses hatred, hostility, or dehumanization toward a specific group or individual, especially those belonging to minority or marginalized communities. • Racial Bias: The post contains biased, discriminatory, or stereotypical statements directed toward one or more racial or ethnic groups. Note: A single post may be labeled as True for multiple bias categories. Information Integrity Classes Each post is also annotated with a single information integrity class, represented by an integer: • -1 → False information (i.e., misinformation or disinformation) • 0 → Unverifiable information (unclear or lacking sufficient evidence) • 1 → True information (verifiable and accurate) Key Notes Versions 1 and 2 are no longer available.本研究提出了面向飓风社会影响的多平台标注数据集（Multiplatform Annotated Dataset for Societal Impact of Hurricane，简称MASH），该数据集包含来自Reddit、TikTok及YouTube的59607条相关社交媒体帖文。 此外，所有相关帖文均采用多模态方法从三大维度进行标注：人道主义类别（Humanitarian Classes）、偏见类别（Bias Classes）与信息完整性类别（Information Integrity Classes），覆盖文本、图像及视频等多模态内容，可为深入分析提供高质量的标注数据集。 据我们所知，MASH是首个大规模、多平台、多模态且多维度标注的飓风相关数据集。我们期望该数据集可服务于飓风社会影响相关研究，例如灾害响应、灾害等级分类、公众情感分析、灾害政策制定以及偏见识别等方向。 使用说明 本数据集包含四个标注文件： • reddit_anno_publish.csv • tiktok_anno_publish.csv • youtube_anno_publish.csv 每个文件均包含帖文ID及对应三大维度的标注结果：人道主义类别、偏见类别与信息完整性类别。 为保护用户隐私，本次仅公开帖文ID。我们建议根据各平台的服务条款，通过其官方API获取完整帖文内容。 - Reddit API (https://www.reddit.com/dev/api) - TikTok API (https://developers.tiktok.com/products/research-api) - YouTube API (https://developers.google.com/youtube/v3) 人道主义类别（Humanitarian Classes） 所有帖文均被标注为7个二元人道主义类别，每个类别的标签取值为： • True：帖文包含该类人道主义信息 • False：帖文未包含该类人道主义信息 该7类人道主义类别具体包括： • 伤亡（Casualty）：帖文报道了飓风期间人员或动物的死亡、受伤或失踪情况。 • 疏散（Evacuation）：帖文描述了因飓风导致的人员或动物疏散、迁移、救援或流离失所情况。 • 设施损毁（Damage）：帖文报道了飓风对基础设施或公共设施造成的损毁。 • 防灾建议（Advice）：帖文提供了与飓风相关的建议、指导或提示，包括如何保障人身安全、保护财产或开展灾前准备。 • 求助请求（Request）：因飓风而寻求帮助、支持或资源的诉求。 • 援助支持（Assistance）：包括个人、社区或组织提供的实物援助以及情绪或心理支持。 • 灾后恢复（Recovery）：帖文描述了飓风后恢复与重建相关的工作或活动。 注意：单条帖文可被多个人道主义类别标注为True。 偏见类别（Bias Classes） 所有帖文均被标注为5个二元偏见类别，每个类别的标签取值为： • True：帖文包含该类偏见信息 • False：帖文未包含该类偏见信息 该5类偏见类别具体包括： • 语言偏见（Linguistic Bias）：帖文包含带有偏见的、不当的或冒犯性的语言，侧重词汇选择、语气或表达方式。 • 政治偏见（Political Bias）：帖文表达了政治意识形态，对特定政治行为体、党派或政策表示支持或反对。 • 性别偏见（Gender Bias）：帖文包含与性别相关的偏见、刻板印象或歧视性语言及观点。 • 仇恨言论（Hate Speech）：帖文包含针对特定群体或个人，尤其是少数群体或边缘群体的仇恨、敌意或非人化表述。 • 种族偏见（Racial Bias）：帖文包含针对一个或多个种族或民族群体的偏见、歧视性或刻板印象言论。 注意：单条帖文可被多个偏见类别标注为True。 信息完整性类别（Information Integrity Classes） 所有帖文同时被标注为单一信息完整性类别，以整数表示： • -1 → 虚假信息（即错误信息或误导性信息） • 0 → 无法验证的信息（表述模糊或缺乏足够证据） • 1 → 真实信息（可验证且准确） 重要说明 版本1与版本2现已不再提供。