@pakhad/locale-en

English locale pack for pakhad. Ships pre-built models — works out of the box, no training required.

What's Included

• Trigram Markov model — 8,650 character trigrams trained on 165,326 English names
• Bloom filter name list — 165,326 names (US Census surnames + SSA baby names), 0.1% false positive rate

Sources: US Census Bureau 2010 Surnames (public domain), Social Security Administration baby names (public domain)

Install

npm install @pakhad/core @pakhad/locale-en

Usage

import { create } from '@pakhad/core';
import en from '@pakhad/locale-en';

const detector = create({ locales: [en] });

detector.detect('John Smith', { fieldType: 'name' });
// { label: "clean", score: 0.075 }

detector.detect('asdfgh qwerty', { fieldType: 'name' });
// { label: "gibberish", score: 0.785 }

Combine with Other Locales

import en from '@pakhad/locale-en';
import indiaLocales from '@pakhad/locale-in';

const detector = create({ locales: [en, ...indiaLocales] });

Model Details

Markov model:  8,650 trigrams, 77 KB
Bloom filter:  165,326 names, 297 KB, 10 hash functions, 0.1% FPR

Building Your Own

If you need a custom English corpus, use @pakhad/train:

npx pakhad-train markov --input your-names.txt --order 3 --output markov.bin
npx pakhad-train namelist --input your-names.txt --fpr 0.001 --output names.bloom

Full Documentation

github.com/nikhilchintawar/pakhad

License

MIT