地球资源数据云——数据资源详情

假新闻和真实新闻数据集 (4K)

发布时间：2026-03-17 14:30:49资源ID：2032007468107075586资源类型：免费

该数据集《Fake & Real News Dataset (4K)》主要用于二分类任务，数据形态以文本为主，应用场景偏向文本内容分析。 题目说明：Structured fake vs real news dataset for classification and NLP modeling. 任务类型：文本二分类。 建议流程：先做文本清洗与分词，再比较 TF - IDF+线性模型 与 预训练语言模型。 评估建议：使用分层切分或交叉验证，优先关注 F1、Recall、AUC 等分类指标。 可用文件：fake_news_dataset_4000_rows.csv。 Description The Fake vs Real News (4K Articles) dataset is a structured and fully cleaned collection of 4000 news articles labeled for binary classification tasks. Each article is categorized as either Fake (0) or Real (1), making this dataset ideal for developing and evaluating fake news detection models. This dataset is specifically designed for Machine Learning and Natural Language Processing (NLP) applications. It includes not only raw text data but also metadata and engineered features that help improve model performance and experimentation flexibility. The dataset is provided in CSV format, contains no missing values, and is ready for direct use in classification pipelines without additional cleaning. Dataset Features

假新闻和真实新闻数据集 (4K)

摘要概览

该数据集《Fake & Real News Dataset (4K)》主要用于二分类任务，数据形态以文本为主，应用场景偏向文本内容分析。 题目说明：Structured fake vs real news dataset for classification and NLP modeling.

任务类型：文本二分类。

建议流程：先做文本清洗与分词，再比较 TF - IDF+线性模型 与 预训练语言模型。

评估建议：使用分层切分或交叉验证，优先关注 F1、Recall、AUC 等分类指标。

可用文件：fake_news_dataset_4000_rows.csv。

Description

The Fake vs Real News (4K Articles) dataset is a structured and fully cleaned collection of 4000 news articles labeled for binary classification tasks. Each article is categorized as either Fake (0) or Real (1), making this dataset ideal for developing and evaluating fake news detection models.

This dataset is specifically designed for Machine Learning and Natural Language Processing (NLP) applications. It includes not only raw text data but also metadata and engineered features that help improve model performance and experimentation flexibility.

The dataset is provided in CSV format, contains no missing values, and is ready for direct use in classification pipelines without additional cleaning.

Dataset Features