ConfigDegenerator

ByJG\TextClassifier\Degenerator\ConfigDegenerator controls the word degeneration behaviour used by the BinaryClassifier spam filter. All setters return $this for fluent chaining.

Parameters

Parameter	Setter	Default	Type	Description
multibyte	setMultibyte(bool)	false	bool	Use multibyte string functions (mb_strtolower, mb_strtoupper, mb_substr) instead of their ASCII counterparts. Required for non-ASCII text.
encoding	setEncoding(string)	'UTF-8'	string	Character encoding for multibyte operations. Only used when multibyte = true.

Usage

ASCII text (default)

use ByJG\TextClassifier\Degenerator\ConfigDegenerator;
use ByJG\TextClassifier\Degenerator\StandardDegenerator;

$degenerator = new StandardDegenerator(new ConfigDegenerator());

Multibyte / Unicode text

$degenerator = new StandardDegenerator(
    (new ConfigDegenerator())
        ->setMultibyte(true)
        ->setEncoding('UTF-8')
);

When to enable multibyte

Enable if your training texts or classified texts contain:

• Accented characters (é, ü, ñ, ö, ...)
• Cyrillic, Greek, Arabic, Hebrew, or other non-Latin scripts
• CJK characters (Chinese, Japanese, Korean)
• Any character outside the ASCII range (0x00–0x7F)

Without multibyte mode, strtolower('Ñoño') does not produce ñoño, which means degenerated variants will be incorrect and case-folding will fail silently for non-ASCII input.

Getters

Method	Returns
isMultibyte()	bool
getEncoding()	string

Related

• Degenerator concepts