Graph World Models: Concepts, Taxonomy, and Future Directions

arXiv cs.AI / 5/1/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes “graph world models” (GWMs) as a unified research paradigm for agent models that represent environments as entities and relations via graph structure rather than flat tensors.
It argues that graph-based world models can mitigate classical world-model issues such as noise sensitivity, error accumulation, and weak reasoning by enabling structured, relational modeling of virtual environments.
The authors introduce a taxonomy of GWMs grounded in relational inductive biases (RIB), splitting approaches into spatial RIB (topological abstraction), physical RIB (dynamic simulation), and logical RIB (causal/semantic reasoning).
For each taxonomy branch, the paper summarizes representative methods and compares their design principles, while also highlighting open problems.
Key future directions include dynamic graph adaptation, probabilistic relational dynamics, multi-granularity inductive biases, and the need for dedicated benchmarks and evaluation metrics for GWMs.

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.