English Hate Speech Posts Superset Merged From Multiple Sources

360,493 posts form a superset of English language content annotated for hate speech. Manueltonneau compiled this collection in April 2024 by merging all publicly available and documented hate speech datasets identified in a systematic survey. The dataset serves as a consolidated resource for text classification research.

Use Cases

• Train binary classifiers to predict the 'hateful' label from post text.
• Analyze linguistic features and patterns within the annotated 'post' text to characterize hate speech.
• Benchmark model performance across a large, consolidated corpus of 360,493 labeled examples.
• Investigate potential label inconsistencies or biases introduced by merging multiple source datasets.

Strengths

• Large scale with 360,493 annotated text posts.
• Consolidates multiple source datasets into a single resource, increasing variety.

Limitations

• Specific column names and data structure are unknown, complicating immediate analysis.
• Merging multiple datasets may introduce label definition inconsistencies and annotation noise.
• Geographic and platform origins of the source data are unspecified, limiting demographic analysis.

Provenance

Source

Merge of all available English hate speech datasets identified via a systematic survey.

Collection Method

Preprocessing and merging of publicly available, documented datasets focusing on hate speech.

Time Range

null

Freshness

Last updated December 2024, with source datasets identified up to April 2024.

Geography

null