Graph World Models: Concepts, Taxonomy, and Future Directions

2026-04-30

AI Summary

Traditional tensor-based world models are prone to noise, suffer from error accumulation, and exhibit limited reasoning capabilities. This work proposes a unified paradigm—Graph World Models (GWM)—that decomposes environments into entity nodes and interaction edges, enabling dynamic system modeling in a structured space. It formally defines GWM for the first time and introduces a tripartite taxonomy grounded in relational inductive biases: spatial, physical, and logical. By establishing a systematic classification framework, this study reviews representative approaches, clarifies design principles and research trajectories, identifies key challenges, and outlines promising future directions, thereby advancing standardization and theoretical foundations in the field.

Abstract

As one of the mainstream models of artificial intelligence, world models allow agents to learn the representation of the environment for efficient prediction and planning. However, classical world models based on flat tensors face several key problems, including noise sensitivity, error accumulation and weak reasoning. To address these limitations, many recent studies use graph structure to decompose the environment into entity nodes and interactive edges, and model virtual environments in a structured space. This paper systematically formalizes and unifies these emerging graph-based works under the concept of graph world models (GWMs). To the best of our knowledge, GWMs have not yet been explicitly defined and surveyed as a unified research paradigm. Furthermore, we propose a taxonomy based on relational inductive biases (RIB), categorizing GWMs by the specific structural priors they inject: (1) spatial RIB for topological abstraction; (2) physical RIB for dynamic simulation; and (3) logical RIB for causal and semantic reasoning. For each model category, we outline the key design principles, summarize representative models, and conduct comparative analyses. We further discuss open challenges and future directions, including dynamic graph adaptation, probabilistic relational dynamics, multi-granularity inductive biases, and the need for dedicated benchmarks and evaluation metrics for GWMs.

Problem

• world models
• graph representation
• relational inductive biases
• structured environment modeling
• reasoning limitations

Innovation

• Graph World Models
• Relational Inductive Biases
• Structured World Representation
• Taxonomy
• Causal Reasoning