ratishsp/fineweb-edu-misinfo

--- license: apache-2.0 task_categories: - text-classification language: - en tags: - misinformation - content-safety - fineweb - pretraining-data - data-quality size_categories: - 100K<n<1M --- # FineWeb-Edu Misinformation Audit A dataset of 200K documents from [FineWeb-Edu](https://huggingface.co/datasets/HuggingFaceFW/fineweb-edu) annotated for misinformation content. FineWeb-Edu uses a Snowflake-arctic-embed embedding model classifier to select "web pages of educational value", but the classifier optimizes for surface-level markers of educational writing (structure, citations, academic tone) and cannot assess whether content is factually accurate or ideologically motivated. ## Dataset composition - **100K documents from known problematic domains** across 7 categories: pseudoscience, climate denial, conspiracy, antivax/medical misinformation, propaganda, hate/extremism, and Holocaust denial - **100K randomly sampled documents** from FineWeb-Edu to establish background rates All documents were annotated by Llama 4 Maverick (meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8) on Together AI for misinformation content. ## Fields | Field | Description | |---|---| | `url` | Original URL from FineWeb-Edu | | `domain` | Domain name extracted from URL (e.g. naturalnews.com) | | `domain_category` | Source domain category (null for random sample) | | `edu_score` | FineWeb-Edu classifier score | | `text` | Full document text | | `llama_label` | Annotation: benign, health_misinfo, pseudoscience, climate_denial, conspiracy_propaganda, hate_extremism | | `llama_confidence` | Annotator confidence (high/medium/low) | | `llama_reason` | Free-text reasoning | ## Key findings - **4.1% of randomly sampled FineWeb-Edu documents** contain misinformation (health misinformation, pseudoscience, conspiracy theories, etc.) - **39% of documents from known problematic domains** contain misinformation content, nearly 10x the overall rate - The full FineWeb-Edu dataset contains 1.53 billion documents. Extrapolating the 4.1% rate, approximately **63 million documents** may contain misinformation. From known problematic domains alone, an estimated **3 million documents** contain confirmed misinformation content. ## Label distribution ### Flagged domains (100K) | Label | Count | |---|---| | benign | 60,804 | | pseudoscience | 16,179 | | health_misinfo | 10,142 | | climate_denial | 5,393 | | conspiracy_propaganda | 4,878 | | hate_extremism | 2,261 | ### Random sample (100K) | Label | Count | |---|---| | benign | 95,865 | | health_misinfo | 2,087 | | pseudoscience | 1,123 | | conspiracy_propaganda | 475 | | hate_extremism | 236 | | climate_denial | 199 | ## Inter-annotator agreement 600 documents (300 flagged, 300 random) were independently annotated by both Llama 4 Maverick and Claude Sonnet 4.6 using the same prompt. Agreement was measured using Cohen's kappa. | Subset | n | Binary agreement | Binary kappa | Multiclass agreement | Multiclass kappa | |---|---|---|---|---|---| | Flagged | 300 | 92.0% | 0.831 | 90.3% | 0.832 | | Random | 300 | 98.7% | 0.850 | 98.0% | 0.779 | | Overall | 600 | 95.3% | 0.862 | 94.2% | 0.842 | Binary confusion matrix (Llama vs Claude, overall): | | Claude: benign | Claude: misinfo | |---|---|---| | **Llama: benign** | 457 | 5 | | **Llama: misinfo** | 23 | 115 | ## Citation If you use this dataset, please cite: ```bibtex @misc{puduppully2026fineweb-edu-misinfo, author = {Puduppully, Ratish}, title = {FineWeb-Edu Misinformation Audit}, year = {2026}, publisher = {HuggingFace}, url = {https://huggingface.co/datasets/ratishsp/fineweb-edu-misinfo} } ```许可证：Apache-2.0 任务类别： - 文本分类 语言： - 英语 标签： - 错误信息 - 内容安全 - FineWeb - 预训练数据 - 数据质量 规模类别： - 10万<n<100万 --- # FineWeb-Edu 错误信息审核数据集 该数据集包含来自FineWeb-Edu的20万条文档，针对错误信息内容进行了标注。FineWeb-Edu采用Snowflake-arctic-embed嵌入模型分类器筛选“具备教育价值的网页”，但该分类器仅针对教育类写作的表层特征（结构、引用、学术语调）进行优化，无法评估内容的事实准确性或意识形态倾向。 ## 数据集构成 - **10万条来自已知问题域名的文档**，涵盖7个类别：伪科学、气候否认论、阴谋论、反疫苗/医疗错误信息、宣传内容、仇恨/极端主义言论以及大屠杀否认论 - **10万条随机采样自FineWeb-Edu的文档**，用于建立基准发生率 所有文档均由Together AI平台上的Llama 4 Maverick（meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8）针对错误信息内容完成标注。 ## 字段说明 | 字段名 | 说明 | |---|---| | `url` | FineWeb-Edu中的原始URL | | `domain` | 从URL提取的域名（例如naturalnews.com） | | `domain_category` | 源域名类别（随机采样样本此字段为空） | | `edu_score` | FineWeb-Edu分类器得分 | | `text` | 完整文档文本 | | `llama_label` | 标注结果：良性、健康类错误信息、伪科学、气候否认论、阴谋论宣传、仇恨极端主义 | | `llama_confidence` | 标注者置信度（高/中/低） | | `llama_reason` | 自由文本标注理由 | ## 核心发现 - **随机采样的FineWeb-Edu文档中，有4.1%包含错误信息**（健康类错误信息、伪科学、阴谋论等） - **来自已知问题域名的文档中，有39%包含错误信息内容**，约为整体发生率的10倍 - FineWeb-Edu完整数据集包含15.3亿条文档。基于4.1%的发生率推算，约**6300万条文档**可能包含错误信息；仅从已知问题域名来看，估计有**300万条文档**包含已确认的错误信息内容。 ## 标签分布 ### 问题域名子集（10万条） | 标签 | 数量 | |---|---| | benign | 60,804 | | pseudoscience | 16,179 | | health_misinfo | 10,142 | | climate_denial | 5,393 | | conspiracy_propaganda | 4,878 | | hate_extremism | 2,261 | ### 随机采样子集（10万条） | 标签 | 数量 | |---|---| | benign | 95,865 | | health_misinfo | 2,087 | | pseudoscience | 1,123 | | conspiracy_propaganda | 475 | | hate_extremism | 236 | | climate_denial | 199 | ## 标注者间一致性 选取600条文档（300条问题域名样本、300条随机样本），由Llama 4 Maverick与Claude Sonnet 4.6使用相同提示词独立完成标注，采用科恩kappa系数评估一致性。 | 子集 | 样本量 | 二元一致性 | 二元kappa系数 | 多分类一致性 | 多分类kappa系数 | |---|---|---|---|---|---| | 问题域名 | 300 | 92.0% | 0.831 | 90.3% | 0.832 | | 随机采样 | 300 | 98.7% | 0.850 | 98.0% | 0.779 | | 整体 | 600 | 95.3% | 0.862 | 94.2% | 0.842 | 整体二元混淆矩阵（Llama 与 Claude 对比）： | | Claude: 良性 | Claude: 错误信息 | |---|---|---| | **Llama: 良性** | 457 | 5 | | **Llama: 错误信息** | 23 | 115 | ## 引用方式 若使用本数据集，请引用： bibtex @misc{puduppully2026fineweb-edu-misinfo, author = {Puduppully, Ratish}, title = {FineWeb-Edu Misinformation Audit}, year = {2026}, publisher = {HuggingFace}, url = {https://huggingface.co/datasets/ratishsp/fineweb-edu-misinfo} }