Download the PHP package bluefission/automata without Composer

BlueFission Automata

The bluefission/automata library is a comprehensive PHP framework designed to facilitate the development and deployment of artificial intelligence systems. It combines classical Good Old-Fashioned AI (GOFAI) techniques with modern machine learning methodologies, providing a multifaceted approach to artificial intelligence. This library is suitable for researchers, developers, and enthusiasts looking to explore and implement AI in their projects.

Features

bluefission/automata integrates a wide array of AI and data science techniques into a cohesive library:

• Expert Systems: Leverage rule-based logic for decision making, simulating the decision-making ability of human experts.
• Game Theory: Analyze competitive environments according to the choices of decision-makers for strategic planning and simulations.
• Scenario Modeling: Inspired by NetLogo, it supports agent-based models for simulating interactions and processes within complex systems.
• Genetic Algorithms: Implement evolutionary algorithms that mimic natural selection for solving optimization problems.
• Monte Carlo Search: Rank candidate actions under uncertainty using repeated seeded rollouts and per-action reward statistics.
• Monte Carlo Tree Search (MCTS): Explore sequential decisions with UCT-style selection, simulation, and backpropagation.
• Path & Graphs: Manage, analyze, and manipulate structures represented graphically including networks of nodes and edges.
• Anomaly Detection: Score behavioral activity, fingerprints, and context to flag unusual or risky patterns.
• Media Ingestion: Normalize text, image, audio, video, document, and URL inputs into consistent pipelines.
• Natural Language Processing (NLP): Tools for text parsing, analysis, and understanding, enabling the library to process and interpret human language.
• Claim Normalization: Normalize raw, wrapped, and adapter-parsed textual claims into inspectable Statement semantics while preserving typed predicates as inert data. See Claim Normalization.
• Bounded Language Prediction: Lightweight Markov and trigram predictors support single-sentence updates and bounded bulk training for moderate local catalogs without requiring a hosted model.
• Large Language Models (LLM): Facilitate prompting and generating responses using large pre-trained models, integrating with tools like GPT for advanced text generation.
• Typed Generation Runs: Describe provider-neutral generation requests, steps, artifacts, diagnostics, partial outcomes, policy, evidence, and adapter-owned execution. See Typed Generation Runs.
• Agent Capabilities: Register deterministic tool contracts, descriptive capability definitions, exact scoped autonomy grants, lifecycle hooks, session memory, Holoscene comprehension, orchestration patterns, DevElation-backed agent state/goal decisions, interpreter-facing integration contracts, and persona orchestration contracts around LLM agent loops. See Agent Persona Orchestration Contracts.
• Adaptive, Deterministic-First Strategy Routing: Select exact, side-effect-free deterministic, learned, or generative strategy adapters under autonomy, eligibility, budget, trace, and explicit escalation policy. Optional Intelligence advice learns contextual quality and efficiency without bypassing those gates. See Strategy Routing.
• LLM Lane Pressure Management: Assess semantic, operational, and execution pressure in provider-neutral agent workflows, with deterministic recommendations and a read-only LLM tool wrapper.
• Feature Engineering: Provides robust tools for transforming raw data into features that better represent the underlying processes to predictive models.
• Data Science: Basic machine learning functionalities alongside data manipulation, preparation, and visualization tools.
• Input Management: Sophisticated input type detection and handling, ensuring that data flows seamlessly through processing pipelines.
• Modular Connectivity: Connect module outputs to other module inputs, creating flexible and dynamic pipeline chains for complex data processing tasks.
• Carrier-Backed Adapters: Normalize runtime state over Develation Arr, Obj, and IData carriers without forcing unrelated modules into one implementation.
• Develation-Native Evaluation Seams: Core orchestration modules now accept DevElation Func evaluators and prefer Num, Arr, and Str helpers for shared numeric, collection, and string handling where that materially improves interoperability.

Using GOFAI and Modern ML Techniques

bluefission/automata uniquely integrates both traditional AI methods and modern machine learning techniques to provide a comprehensive toolkit:

• GOFAI Techniques: The library utilizes symbolic AI methods for creating systems that reason with logic and predefined rules, suitable for scenarios where decision paths need to be transparent and based on human-like logic.
• Modern Machine Learning: Incorporates statistical learning techniques for pattern recognition, predictive modeling, and data-driven decision-making, allowing the system to adapt and learn from data.

Getting Started

To get started with bluefission/automata, clone the repository and include it in your PHP project:

Monte Carlo examples:

Language prediction example:

Contributing

We welcome contributions to the bluefission/automata project. Please read the CONTRIBUTING.md for details on our code of conduct and the process for submitting pull requests to us.

Shared contributor guidance for ecosystem boundaries, dependency notes, public issue hygiene, coordination, and evidence expectations lives in Automata Ecosystem Boundaries.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.