Fixed 20 Newsgroups Text Classification Dataset

A corrected version of the 20 Newsgroups dataset, processed using the Galileo ML Data Intelligence Platform to fix errors. It is intended for natural language experimentation and benchmarking. The dataset is monolingual English, crowdsourced, and categorized for multi-class topic classification.

Use Cases

• Train a text classification model for the 20 newsgroup topics using the corrected text data.
• Benchmark topic classification algorithms on the fixed dataset to measure performance improvements over the original version.
• Analyze the distribution of text across the 20 newsgroup categories to understand data composition.

Strengths

• Dataset size is categorized as between 10K and 100K entries.
• Data has been processed to fix a multitude of errors from the original version.
• Specifically formatted for the task of multi-class text classification.

Limitations

• The exact number of rows, columns, and specific error corrections are not detailed in the provided input.
• The dataset's curation and correction methodology is only summarized, with full details requiring external review.
• As a fixed version of a classic dataset, it may inherit any inherent biases or limitations from the original 20 Newsgroups collection.

Provenance

Source

Original 20 Newsgroups dataset, corrected by galileo-ai.

Collection Method

Crowdsourced, with errors fixed using the Galileo ML Data Intelligence Platform.

Freshness

Last updated on 2022-10-25.

Geography

Region is indicated as US.