An MAB-ST–Based Relational Framework for Game AI and Simulation (Runnable Python Appendix)
Year: 2026
Type: article
Abstract: This material is the practical, runnable Python appendix of the MAB-ST (Markó Anna Beáta – Set Theory) framework.Its purpose is to demonstrate, through concrete, working code examples, how the MAB-ST relational world model can be applied in interactive systems, particularly in game AI and simulation contexts. This appendix is not a complete game engine nor a production-ready framework. Instead, it provides a minimal, clean, educational reference implementation that illustrates: how a “world” can be represented as a network of relations between entities, how behavior can be derived from relational patterns through a meta-level rule layer, how the system can be extended with new subsystems (e.g., reputation, factions, quests) without fragmenting the overall control logic. The package contains fully runnable Python files (no external dependencies) that build on each other to demonstrate: a minimal relational “world core” implementation, a relational NPC behavior model (the “Town Square” scenario), a classical state-machine-based contrast implementation for comparison, relational extensions (reputation, faction influence, loyalty, quest logic), and a lightweight validation and testing layer. Important reference and prerequisite:This Python appendix is built on the theoretical “core” of the MAB-ST framework. The conceptual foundations and licensing terms of MAB-ST are available under the following DOI: MAB-ST Core / Backbone / Read This First / License DOI: 10.5281/zenodo.18518738 Users are strongly advised and expected to read the documents available under the above DOI before using this appendix, as they contain: the theoretical foundation of MAB-ST, the conceptual framework and terminology, and the applicable licensing conditions. This material does not require prior knowledge of the full mathematical formalization of MAB-ST, but without reading the “core” documents, the underlying modeling paradigm and usage framework may be misunderstood.The code examples are fully runnable and suitable for: research, educational, and experimental purposes, exploring relational modeling patterns in interactive systems, and gaining practical insight into the application of the MAB-ST approach. ⚠️ License notice:This material may be used freely for research and educational purposes.Any industrial, commercial, or product-integrated use requires a separate license from the rights holder of MAB-ST. Target Audience and Use Cases: What Is ThisAppendix For?Explanatory Appendix: How to Interpret the MAB–STFrameworkThis appendix clarifies the intended interpretation of the MAB–ST Framework.Its purpose is to resolve common misunderstandings that typically arise whena new meta-level theoretical framework is introduced, especially in the contextof software engineering, artificial intelligence, and system design.The Role of MAB–STMAB–ST is not a concrete implementation technology but an abstract, meta-level framework for thinking and modeling. Its role is not to provide ready-madesolutions to specific problems, but to offer a unified conceptual structure fordescribing and reasoning about complex systems.MAB–ST:• is not a programming language,• is not an executable engine,• is not a ready-made AI architecture,• but a theoretical framework within which world models, relations, andrule systems can be described and reasoned about in a consistent manner.Meta-Level Structure and the Question of RigidityEvery meta-level framework faces the question of “rigidity”: the risk thatthe framework becomes too restrictive in what it can express.1MAB–ST deliberately adopts a strong structural foundation (the worldquadruple: state, relation, kernel, rule), which acts as a conceptual pre-filter. What cannot be meaningfully expressed within this structure is notconsidered a well-formed world description within the framework.This form of “rigidity” is not a technical limitation but a philosophicaldesign choice: the goal is to prevent the emergence of inconsistent modelsburdened with hidden contradictions. At the same time, the frameworkdoes not prescribe how concrete subsystems (AI, reputation, quests, econ-omy) must be implemented.Common MisconceptionsThe following statements reflect frequent but misleading interpretationsof the MAB–ST framework:– “MAB–ST is a data structure.”No, it is an abstract conceptual framework for thinking and modeling.– “This replaces classical AI architectures.”No, it is compatible with and integrable into classical AI architec-tures.– “This is a ready-made engine.”No, it is a theoretical foundation accompanied by demonstrationalimplementations.Relation to Classical ArchitecturesMAB–ST does not replace classical design patterns (state machines, be-havior trees, ECS, rule-based systems), but organizes them at a higherlevel of abstraction into a unified world model.In practical implementations:– classical architectures remain fully usable,– MAB–ST provides a unifying conceptual layer for integrating theminto a shared world model,– relational world modeling enables loose coupling between subsystems.What Does the Python Appendix Demonstrate?The Python appendix does not “prove” the mathematical validity of MAB–ST. Instead, it demonstrates that:– relation-based world modeling can be implemented at the level ofpractical code,2– meta-level rule modeling can be illustrated with simple, transparentimplementations,– multiple subsystems (AI, reputation, factions, quests) can be inte-grated into a unified world structure.This appendix is therefore not a “product” but a didactic bridge betweenthe theoretical framework and practical software design.Intended AudienceThis appendix is intended for:– software engineers and system architects exploring alternative world-modeling paradigms,– AI researchers interested in relational and meta-level modeling ap-proaches,– developers of simulation, game AI, or complex interactive systems,– researchers investigating formal frameworks for consistent systemmodeling.The material assumes basic familiarity with software design concepts, butdoes not require prior knowledge of the formal MAB–ST theoretical foun-dations.What Is the “Core Document”?The MAB–ST core document provides the theoretical and conceptualfoundation of the framework. It defines:– the structure of the world quadruple (state, relation, kernel, rule),– what counts as a “world description” within the framework,– and along which conceptual constraints a model is considered coher-ent and interpretable.The core document is not a tutorial and not an implementation guide. Itspurpose is not immediate usability, but to provide a stable interpretativefoundation for all later applications and extensions.Official reference of the MAB–ST core document:MAB–ST Core / Read This First / License (DOI: 10.5281/zen-odo.18518738).The Python appendix and all further extensions are conceptually built onthis core document.3How to Read the Material (Conceptual Map)The material is structured in layers. Readers are not expected to startwith the core document in order to run the code, but understanding theconceptual intent of the framework is strongly supported by it.– The Core (Gerinc) defines the conceptual frame.– The Python Appendix demonstrates how relational world model-ing and meta-level rules can be expressed in practical code.– The Extensions (reputation, factions, quests, validation) illustratehow new subsystems can be added without restructuring the entiresystem.The Python appendix should be read as a didactic bridge between theabstract framework and practical software design. It shows how one canthink and structure code within the MAB–ST perspective, not how tobuild a finished engine.Why This Is Structured Differently Than Classical Ap-proachesClassical AI and game architectures often distribute logic across multiplecomponents, state machines, and event handlers. The relational approachdemonstrated here differs in that:– world state is represented centrally as relations,– behavior emerges from patterns in the world model,– new systems are primarily added by introducing new relations andrules, not by expanding control-flow meshes.This difference is conceptual rather than technological: MAB–ST does notreplace existing architectures, but offers a unifying modeling perspectiveabove them.
Author: Markó, Anna Beáta
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7131103774
FWCI:
Cites:
Cited by:
Related to: