Legal eDiscovery Text Data for Classification Tasks

A text dataset for legal eDiscovery classification tasks, created by Cleanlab and last updated on Hugging Face in January 2026. It includes a main file and a full version, but the specific number of documents and detailed contents are not provided in the description.

Use Cases

• Classifying legal documents for relevance in eDiscovery based on text content.
• Training text classifiers to identify privileged or confidential information in legal corpora.
• Benchmarking NLP models on domain-specific legal language understanding tasks.

Strengths

• Dataset is hosted on the Hugging Face platform, which provides a standard interface for access.
• Two file versions (main and full) are available, suggesting options for different scale requirements.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Cleanlab

Freshness

Last updated 2026-01-20 02:35:16; freshness should be verified.