Work

Dynamic Graph Grammars for Generating Complex System Models

Year: 2026
Type: preprint
Abstract: This paper proposes a novel approach to system modeling using dynamic graph grammars. Traditional graph grammars are inherently static, generating models based on pre-defined rules without any adaptive behavior. This research introduces a framework where graph grammar rules can dynamically adjust their execution based on observed system states, creating models that are more realistic and responsive to evolving environments. The core of this approach involves integrating "watch" clauses – rules triggered only when specific conditions are satisfied within the graph. Furthermore, the paper explores the use of machine learning, specifically reinforcement learning, to optimize the configuration of these watch clauses, learning the most effective strategies for adapting the graph grammar to complex system dynamics. This dynamic adaptation significantly expands the capabilities of graph grammars, enabling the generation of models capable of not just representing systems, but actively responding to and evolving within them. The key innovation lies in combining the declarative power of graph grammars with the adaptive learning capabilities of machine learning.
Language: English
Subfield: Software
Open Access status: green
OpenAlex ID: W7204576786
Cites:
Cited by:
Related to: 10