Patent Document Clustering Benchmark With Nine Categories

Test set for clustering documents from the Big Patent dataset, containing documents belonging to nine distinct categories. It is part of the Massive Text Embedding Benchmark (MTEB) for evaluating embedding models on legal and written domain text.

Use Cases

• Evaluate embedding models on clustering performance across nine patent document categories.
• Benchmark text representation methods for legal and intellectual property documents.
• Analyze document similarity within and between the nine specified patent categories.

Strengths

• Documents are categorized into nine distinct classes, providing a structured clustering target.
• Part of the standardized Massive Text Embedding Benchmark (MTEB) for model evaluation.
• Focuses on the legal and written domains, specifically patent documents.

Limitations

• Dataset size, row count, and specific column features are unknown.
• Only contains a test set, limiting its use for training or validation tasks.
• The temporal coverage and geographic origin of the patent documents are unspecified.

Provenance

Source

Derived from the Big Patent dataset, hosted on Hugging Face by mteb.

Collection Method

Clustering of documents from the Big Patent dataset, curated for the MTEB benchmark.

Time Range

null

Freshness

null

Geography

null