BlogCorpus

A huge dataset of blog posts from across the internet. Contains over 60M posts, along with supporting metadata like author_id and labels, and engagement data like number of comments and likes.

For more, see Primordic

Description

This dataset contains blog posts scraped from various blog providers including wordpress and blogger.

Uses

• Text Generation
• Text Summarization
• Text Categorization

Dataset Structure

Each row represents a blog post from a particular blog.

Columns are defined as follows:

• blog_id - a unique ID identifying the blog that this post came from
• author_id - a unique ID identifying the author of the blog. Usually all posts in a given blog are from the same author. Exact name has been redacted for privacy.
• published - the date that the blog post was published
• title - the title of the post
• body - the main text content of the post. Can either be plain text or HTML depending on the blog
• labels - some topic labels provided by the author. Only available for some blogs.
• comments - the number of comments on the post. 0 means no comments, None means comments were not enabled.
• likes - the number of likes on the post. 0 means no likes, None means likes were not enabled.
• slug - the URL slug for the page

Bias, Risks, and Limitations

Data comes directly from the blogs without any filtering. Some blog posts may contain adult content or harmful themes.

More Information

For more, see Primordic