fingerprints

A library to generate entity fingerprints.


Keywords
names, people, companies, normalisation, iso20275, clustering, company-data, deduplication, entity, fingerprint, nlp
License
MIT
Install
pip install fingerprints==1.2.3

Documentation

fingerprints

package

This library helps with the generation of fingerprints for entity data. A fingerprint in this context is understood as a simplified entity identifier, derived from it's name or address and used for cross-referencing of entity across different datasets.

Usage

import fingerprints

fp = fingerprints.generate('Mr. Sherlock Holmes')
assert fp == 'holmes sherlock'

fp = fingerprints.generate('Siemens Aktiengesellschaft')
assert fp == 'ag siemens'

fp = fingerprints.generate('New York, New York')
assert fp == 'new york'

Company type names

A significant part of what fingerprints does it to recognize company legal form names. For example, fingerprints will be able to simplify Общество с ограниченной ответственностью to ООО, or Aktiengesellschaft to AG. The required database is based on two different sources:

Wikipedia also maintains an index of types of business entity.

See also

  • Clustering in Depth, part of the OpenRefine documentation discussing how to create collisions in data clustering.
  • probablepeople, parser for western names made by the brilliant folks at datamade.us.