Assessing Vaccine-Related Content for Journalistic Quality: A large-scale dataset and article repository

This dataset was produced through a collaboration with the NSF-funded ARTT project (led by Hacks/Hackers) and Overtone. It consists of 1,000 vaccine-related articles, pulled from a wide variety of news media sources, with associated scores based on their journalistic quality. The scores were provided through Overtone’s algorithm, and range from one (low-quality or low informational value add) to five (high-quality or high informational value add). Articles were sourced from traditional journalism outlets (news and news-leaning websites), as well as non-journalistic sources of vaccine information, such as governmental websites, healthcare and NGO websites, and medical journals. Given the algorithm’s focus on editorial content, as opposed to other metrics such as author, outlet, or engagement, analyzing a diverse set of article types allowed the research team to examine how different styles of vaccine-related content measured against traditional journalistic quality standards.<strong> </strong>Therefore, this dataset provides a unique insight into the spectrum of vaccine reporting, and serves as a contribution to the field of automated quality assessment. <strong>About ARTT:</strong> The Analysis and Response Toolkit for Trust (ARTT) project is focused on helping people engage in trust-building ways when discussing vaccine efficacy and other topics online. <strong>About Overtone:</strong> Overtone has built a Natural Language Processing algorithm that finds and sorts online content by its intrinsic qualities, rather than clicks or shares. Their AI assesses texts for journalistic signals that demonstrate human effort. For any questions about this dataset, please contact [e-mail].本数据集由与美国国家科学基金会（National Science Foundation，简称NSF）资助的ARTT项目（由Hacks/Hackers主导）及Overtone合作制作。该数据集包含1000篇与疫苗相关的文章，取材自各类新闻媒体来源，并附带基于其新闻采编质量的评分。评分由Overtone的算法生成，取值区间为1至5分：1分代表质量偏低或信息增值价值有限，5分代表质量上乘或信息增值价值突出。文章来源既包括传统新闻媒体机构（新闻网站及偏向新闻的平台），也涵盖非新闻类的疫苗信息渠道，例如政府网站、医疗与非政府组织（NGO）网站以及医学期刊。鉴于该算法聚焦于编辑内容本身，而非作者、发布平台或互动量等其他指标，研究团队通过分析多样化的文章类型，得以检验不同风格的疫苗相关内容是否符合传统新闻采编质量标准。因此，本数据集为全面洞察疫苗报道的全貌提供了独特视角，同时也为自动化质量评估领域贡献了研究资源。**关于ARTT项目：** 信任分析与响应工具包（Analysis and Response Toolkit for Trust，简称ARTT）项目致力于帮助人们在网络上讨论疫苗有效性及其他话题时，以增进信任的方式开展交流。**关于Overtone：** Overtone开发了一款自然语言处理（Natural Language Processing）算法，可基于内容的内在属性而非点击量或分享量来识别并分类网络内容。其人工智能系统会对文本进行评估，以识别体现人类采编投入的新闻采编信号。若对本数据集有任何疑问，请联系[e-mail]。