Download the PHP package byjg/text-classifier without Composer
On this page you can find all versions of the php package byjg/text-classifier. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
Download byjg/text-classifier
More information about byjg/text-classifier
Files in byjg/text-classifier
Package text-classifier
Short Description A PHP text classifier supporting binary spam filtering (Robinson-Fisher Bayesian) and multi-class Naive Bayes classification, with optional LLM-assisted active learning fallback.
License
Homepage https://github.com/byjg/b8
FAQ
Example:
In general, it is recommended to use always a project to download your libraries. In an application normally there is more than one library needed.
- Some hosting areas are not accessible by a terminal or SSH. Then it is not possible to use Composer.
- To use Composer is sometimes complicated. Especially for beginners.
- Composer needs much resources. Sometimes they are not available on a simple webspace.
- If you are using private repositories you don't need to share your credentials. You can set up everything on our site and then you provide a simple download link to your team member.
- Simplify your Composer build process. Use our own command line tool to download the vendor folder as binary. This makes your build process faster and you don't need to expose your credentials for private repositories.
Informations about the package text-classifier
sidebar_key: text-classifier tags: [php, text-classification, ai]
text-classifier — Bayesian Text Classifier
A PHP library for statistical text classification. Provides two independent engines:
- BinaryClassifier — Binary Robinson-Fisher Bayesian filter. Classifies text as spam or ham. Designed for high-accuracy two-class filtering with word degeneration support.
- NaiveBayes — Multi-class Naive Bayes classifier. Classifies text into any number of user-defined categories. Suitable for language detection, topic tagging, content routing, and similar tasks.
Both engines return a ClassificationResult with the winning category, confidence score, and all category scores. Both support optional LLM injection for automatic escalation when the statistical model is uncertain — the LLM decision is fed back as training data, improving the model over time (active learning).
Both engines share the same tokenisation pipeline (StandardLexer, StandardDegenerator) and support pluggable storage backends (in-memory, SQLite, MySQL, PostgreSQL, GDBM).
Installation
Requires PHP >=8.3. The GDBM storage backend additionally requires ext-dba.
Quick Example
Spam filter:
Multi-class classifier:
Documentation
| Section | Description |
|---|---|
| Getting Started | Installation, requirements, first working example |
| Guides: Spam Filter | Training, classifying, choosing storage |
| Guides: Multi-class | Training categories, classifying, persistence |
| Guide: LLM-Assisted Classification | Automatic LLM fallback and active learning |
| Concepts | How the algorithms work, architecture overview |
| Reference | Full API, configuration parameters, error codes |
Acknowledgements
This library is inspired by the original b8 spam filter written by Tobias Leupold. The core algorithm, Robinson-Fisher probability model, token degeneration approach, and the tc* internal variable convention all originate from his work. This project modernises the codebase for PHP 8.3+, replaces the storage layer with byjg/micro-orm and byjg/migration, and adds a multi-class NaiveBayes engine built on the same tokenisation pipeline.
Dependencies
All versions of text-classifier with dependencies
byjg/migration Version ^6.0
byjg/llm-api-objects Version ^6.0
openai-php/client Version ^0.19
php Version >=8.3 <8.6
symfony/http-client Version ^7.4|^8.0
nyholm/psr7 Version ^1.8