Evan823/headlines_clean_2022_2025_slugs

一个轻量级的**Fox News与NBC News**二分类文本数据集，基于**新闻URL**构建。每个样本使用从URL最后一个有意义的路径组件（slug）派生并规范化为简短、干净文本的**伪标题**，而非完整文章文本。这使得数据集便于快速基线测试（TF-IDF + 线性模型）和稳定的Transformer微调。 - **任务**：二分类文本分类（预测发布者/来源） - **标签**：`0 = NBC`，`1 = Fox` - **文本**：规范化的URL slug / 伪标题 - **语言**：英语 - **许可证**：MITA lightweight **Fox News vs. NBC News** binary text classification dataset built from **news URLs**. Instead of using full article text, each example uses a **pseudo-headline** derived from the URL’s last meaningful path component (slug) and normalized into short, clean text. This makes the dataset convenient for fast baselines (TF-IDF + linear models) and also stable for transformer fine-tuning. - **Task**: binary text classification (predict publisher/source) - **Labels**: `0 = NBC`, `1 = Fox` - **Text**: normalized URL slug / pseudo-headline - **Language**: English - **License**: MIT