README.md

metadata

language:
  - en
license: mit
pretty_name: EUIPO Trademark Nice Classification Terms
size_categories:
  - 10K<n<100K
tags:
  - trademark
  - classification
  - nice-classification
  - euipo
  - intellectual-property
  - rag
  - retrieval
task_categories:
  - text-classification
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/euipo.parquet
dataset_info:
  features:
    - name: id
      dtype: string
    - name: nice_class
      dtype: int64
    - name: description
      dtype: string
    - name: approval_type
      dtype: string
  splits:
    - name: train
      num_examples: 87752

EUIPO Trademark Nice Classification Terms

Dataset Description

This dataset contains 87,752 goods and services terms classified under the Nice Classification system, as catalogued by the European Union Intellectual Property Office (EUIPO) via their TMclass tool.

The data was scraped from TMclass in early 2025.

Source

• Origin: TMclass – EUIPO
• Method: Web scraping
• Date Collected:

Dataset Structure

Schema

Column	Type	Description
id	string	Unique identifier for the term (e.g. en/1/262)
nice_class	int64	Nice Classification class number (1–45)
description	string	The goods or services term description (e.g. "Absorbent expandable graphite")
approval_type	string	Approval status: "M" (MGS — harmonised), "V" (validated by national office), or "" (unspecified)

Data Sample

[
  {"id": "en/1/262",       "nice_class": 1, "description": "Abrasives (Auxiliary fluids for use with -)", "approval_type": ""},
  {"id": "en/1/234227210", "nice_class": 1, "description": "Absorbent expandable graphite",              "approval_type": "V"},
  {"id": "en/1/616833",    "nice_class": 1, "description": "Absorbent graphites",                        "approval_type": "M"}
]

Splits

Split	Examples
train	87,752

Nice Classification Overview

The Nice Classification is an international system for classifying goods and services for trademark registration. It comprises 45 classes — classes 1–34 cover goods and classes 35–45 cover services.

Use Cases

• Training text classifiers to predict Nice class from product/service descriptions
• Building search and recommendation systems for trademark registration
• Semantic analysis of goods and services terminology
• Augmenting NLP pipelines with structured IP classification data

Limitations

• The dataset is a snapshot from a specific point in time and may not reflect the latest TMclass updates
• Only English-language terms are included
• Some approval_type values are empty, indicating unspecified status

Citation

@dataset{euipo_trademark_nice,
  title   = {EUIPO Trademark Nice Classification Terms},
  author  = {hashiromer621@gmail.com},
  year    = {2026},
  note    = {Scraped from TMclass (https://tmclass.tmdn.org/)}
}

License

This dataset is released under the MIT License.