Executable Ontologies for Game Development: From Static Models to Living Game Worlds

Modern game development relies heavily on static data models, hard-coded logic, and tightly coupled systems. While this approach works for predefined mechanics, it struggles when games need to evolve dynamically, support emergent behavior, or adapt to player actions in real time.

In this article, Alexander Boldachev explores how executable ontologies can be used as a foundation for game development. Instead of treating ontologies as passive classification schemes, this approach turns them into living, executable models that define not only game entities and relationships, but also rules, constraints, and causal dynamics.

The paper introduces an event-driven semantic architecture where game logic is expressed as executable models rather than imperative code. Game objects, actions, and processes are represented as streams of semantically typed events, allowing game worlds to evolve organically over time. This makes it possible to modify mechanics, add new behaviors, and extend game logic without rewriting core systems or breaking existing gameplay.

Key topics covered in the article include:

• Why traditional object-centric architectures limit emergent gameplay

• How executable ontologies unify data, logic, and game rules

• Event-driven semantic modeling of actions, states, and processes

• Dynamic modification of game mechanics at runtime

• Implications for AI-driven agents, simulations, and open-ended worlds

This approach is especially relevant for complex simulations, sandbox games, strategy games, and AI-driven environments where adaptability, traceability, and semantic clarity are critical.

The full academic paper is available on arXiv: https://arxiv.org/abs/2601.07964