news-media-bias.data.json

The prevalence of bias in the news media has become a critical issue, affecting public perception on a range of important topics such as political views, health, insurance, resource distributions, religion, race, age, gender, occupation, and climate change. The media has a moral responsibility to ensure accurate information dissemination and to increase awareness about important issues and the potential risks associated with them. This highlights the need for a solution that can help mitigate against the spread of false or misleading information and restore public trust in the media.<b>Data description</b>This is a dataset for news media bias covering different dimensions of the biases: <i>political, hate speech, political, toxicity, sexism, ageism, gender identity, gender discrimination, race/ethnicity, climate change, occupation, spirituality, </i>which makes it a unique contribution.The dataset used for this project <b>does not contain any personally identifiable information (PII)</b>.<b>Data Format:</b>- ID: Numeric unique identifier.- Text: Main content.- Dimension: Categorical descriptor of the text.- Biased_Words: List of words considered biased.- Aspect: Specific topic within the text.- Label: Bias True/False value- Aggregate Label: Calculated through multiple weighted formulae<br><b>Annotation Scheme:</b>1. Bias Label: Indicate the presence/absence of bias (e.g., no bias, mild, strong).2. Words/Phrases Level Biases: Identify specific biased words/phrases.3. Subjective Bias (Aspect): Capture biases related to content aspects.<br><b>Annotation Process:</b>Manual Labeling --&gt; LLM based labelling --&gt;Semi-Supervised Learning --&gt; Human Verifications (iterative process)<br>The scheme employs a mix of manual labeling, GPT-based labeling, human verification, and semi-supervised learning for refined and accurate annotation.We want to offer open and free access to dataset, ensuring a wide reach to researchers and AI practitioners across the world. The dataset should be user-friendly to use and uploading and accessing data should be straightforward, to facilitate usage.<br>

新闻媒体中偏见的泛滥已成为一项关键议题，其影响覆盖政治观点、健康、保险、资源分配、宗教、种族、年龄、性别、职业及气候变化等诸多重要领域的公众认知。媒体肩负道德责任，需保障信息的准确传播，并提升公众对重要议题及其潜在风险的认知水平。这一现状凸显出亟需一种解决方案，以助力遏制虚假或误导性信息的扩散，重塑公众对媒体的信任。<b>数据集说明</b>本数据集针对新闻媒体偏见问题，涵盖多维度偏见类型，包括<i>政治偏见、仇恨言论、政治偏见、毒性言论、性别偏见、年龄歧视、性别认同、性别歧视、种族/族裔、气候变化、职业、精神信仰</i>，这使得本数据集具备独特的研究价值。本项目所使用的数据集<b>未包含任何个人可识别信息（PII）</b>。<b>数据格式：</b>- 编号（ID）：数字型唯一标识符。- 文本（Text）：新闻主体内容。- 维度（Dimension）：文本的分类描述项。- 偏见词汇（Biased_Words）：被判定为带有偏见的词汇列表。- 主题（Aspect）：文本所涉及的特定议题。- 标签（Label）：表示偏见存在与否的布尔值（True/False）。- 聚合标签（Aggregate Label）：通过多重加权公式计算得到的结果。<b>标注方案：</b>1. 偏见标签：标注偏见的存在与程度（例如：无偏见、轻度偏见、重度偏见）。2. 词汇/短语级偏见识别：定位文本中具体的偏见性词汇或短语。3. 主题维度主观偏见：捕捉与内容主题相关的偏见倾向。<b>标注流程：</b>人工标注 → 基于大语言模型（LLM）的标注 → 半监督学习 → 人工校验（迭代式流程）。本标注方案融合了人工标注、基于GPT的标注、人工校验以及半监督学习技术，以实现精细化且精准的标注效果。我们致力于向全球范围内的研究者与人工智能从业者开放并免费提供本数据集，以保障其广泛的应用触达。本数据集应具备良好的易用性，数据上传与访问流程应简洁直观，以便于使用者高效开展相关研究工作。