Data-Driven Discovery of Graph Grammars from Complex Network Data
Year: 2026
Type: preprint
Abstract: This paper investigates the feasibility of using data-driven techniques to automatically discover graph grammars from complex network data. Traditional graph grammar formalisms require manual definition of rules, a process that can be both time-consuming and dependent on expert knowledge. This research proposes a novel approach leveraging machine learning algorithms to identify recurring structural patterns within networks and subsequently construct graph grammars that reflect these patterns. The core claim is that algorithms can accurately represent the evolution of complex networks by discovering these grammars directly from observational data. We explore the application of unsupervised learning methods, specifically clustering techniques, to identify dominant structural motifs. These motifs are then used as the basis for generating a graph grammar, allowing for the automated description of network dynamics. The primary contribution lies in the automation of this discovery process, potentially unlocking new insights into the complex and often unpredictable behavior of networks across diverse domains. The methodology presented provides a framework for understanding and modeling network evolution without relying on human-defined rules.
Author: Zhang, Jincheng
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204946556
