Do LLM language abilities generalize to graph-structured social networks?

Determine whether large language models (LLMs) generalize to graph-structured objects by establishing whether they can generate social networks that reproduce key structural characteristics of real social networks, such as low edge density and long-tailed degree distributions.

Background

The paper motivates using LLMs for zero-shot, flexible generation of social networks, contrasting them with classical random graph models and deep generative models that either lack realism or require extensive training data. Despite LLMs’ success in simulating human interactions, prior work shows they can struggle with graph reasoning tasks, raising uncertainty about whether their natural language capabilities extend to structured graph objects.

This uncertainty is pivotal for applications like epidemiological modeling and social simulation, where generated networks must plausibly match real-world structural properties (for example, sparsity and heavy-tailed degree distributions). The authors frame this as a central research question driving their study of prompting methods for LLM-based social network generation.

References

On the other hand, LLMs sometimes struggle with reasoning over graphs and it is unclear if their language abilities generalize to structured objects like networks, so that they can reproduce structural characteristics of social networks such as low density and long-tailed degree distributions.

— LLMs generate structurally realistic social networks but overestimate political homophily  (2408.16629 - Chang et al., 2024) in Section 1 (Introduction)

Existing benchmarks, therefore, often report narrow fidelity metrics and leave open the question of whether a generated graph merely memorizes the reference or remains useful for downstream tasks.

— Scalable Hierarchical Graph Generation via Soft Community Structure  (2610.12163 - Tüzen et al., 8 Oct 2026) in Introduction, paragraph beginning “Evaluation in this setting is itself underdeveloped.”