bonjour

[!CAUTION]

Work in progress, first usable version hasn't landed yet!

Experimental crate to extract first names with a confidence level.

Expected output may be something like this:

{
  "input": "Quentin Richert",
  "first_name": "Quentin",
  "confidence": 0.95,
  "gender": "male",
  "country": "FR"
}

The idea is that is also "detects", or at lease significantly reduces confidences in company names, for instance:

// The company marker 'SAS' significantly reduces confidence.
{
  "input": "Quentin Richert SAS",
  "first_name": "Quentin",
  "confidence": 0.1,
  "gender": "male",
  "country": "FR"
}

Up to no detection at all:

{
  "input": "Les Motards d'Alsace",
  "first_name": null,
  "confidence": 0.0,
  "gender": null,
  "country": null
}

Country and gender hints

Gender is not a property of a name alone — Simone is female in France, male in Italy. Pass the user's country and/or gender as hints and they resolve each other: a country pins the gender, a gender pins the country.

$ bonjour --country=IT Simone Veil
{
  "input": "Simone Veil",
  "first_name": "Simone",
  "confidence": 0.65,
  "gender": "male",
  "country": "IT"
}

$ bonjour --country=FR Simone Veil
{
  "input": "Simone Veil",
  "first_name": "Simone",
  "confidence": 0.7,
  "gender": "female",
  "country": "FR"
}

With no hint and a name whose gender differs by country, gender is left null rather than guessed:

{
  "input": "Simone",
  "first_name": "Simone",
  "confidence": 0.7,
  "gender": null,
  "country": "FR"
}

License

The source code is available under the 0BSD license.

Datasets distributed with or used to build this project are compiled from publicly available information. They are not covered by the 0BSD license; their contents remain subject to any applicable rights and source terms.