Just for FUNS: LLM-Guided Spatio-Temporal Graph Node Generation for Forecasting Unobserved Node States
Abstract
Spatio-temporal forecasting is a cornerstone of logistics, urban planning, and intelligent transportation systems. However, constrained by deployment costs and maintenance resources, sensor networks often lack comprehensive spatial coverage, rendering Forecast Unobserved Node States (FUNS) a critical yet formidable challenge. Conventional models rely on historical observations and typically falter when encountering nodes without prior records. To address this, we redefine the problem as a conditional generation task on spatio-temporal graphs and propose GenST, a framework that introduces Large Language Models (LLMs) as a semantic bridge, leveraging a pre-trained LLM fine-tuned to extract rich semantic features from node descriptions, such as functional zones and road network structures, to compensate for missing spatio-temporal signals. Specifically, we design a two-stage generative architecture: a Spatio-Temporal VAE first compresses spatio-temporal dynamics into a latent space, followed by a Generative Transformer (GenT) that reconstructs the future states of unobserved nodes from noise, guided by multi-modal conditions including semantics, geographic coordinates, and neighborhood contexts. Experiments on six traffic and two non-traffic datasets show GenST significantly outperforms existing baselines in zero-shot prediction tasks, demonstrating the practical potential of semantic-guided generation for mitigating spatio-temporal data sparsity.
Related
- How Much Graph Is Too Much? Graph Construction and Exposure in Spatio-Temporal Forecasting
- STCFormer: Adaptive Spatio-Temporal Modeling with Dynamic Cluster Transformer for Station-based Weather Forecasting
- Short-term metro passenger flow forecasting via decoupled spatio-temporal modeling with a diffusion-convergence graph and Swin-Enhanced LSTM
- A Non-Stationary Spatio-Temporal Covariance Model with Dynamic Advection Effects for Rainfall Data
- A case study of hybrid statistical models for analyzing spatio-temporal effects in historical map series of Donau-Ries county
BibTeX
@misc{li2026just,
title = {{Just for FUNS: LLM-Guided Spatio-Temporal Graph Node Generation for Forecasting Unobserved Node States}},
author = {Shuhao Li and Weidong Yang and Changan Liu and Wei Zhuo and Yingbo Zhou and Fan Zhang and Siqiang Luo},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.08818},
eprint = {2610.08818},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.08818},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Just for FUNS: LLM-Guided Spatio-Temporal Graph Node Generation for Forecasting Unobserved Node States
AU - Shuhao Li
AU - Weidong Yang
AU - Changan Liu
AU - Wei Zhuo
AU - Yingbo Zhou
AU - Fan Zhang
AU - Siqiang Luo
PY - 2026
DA - 2026/10/08
PB - arXiv
DO - 10.48550/arxiv.2610.08818
UR - https://arxiv.org/abs/2610.08818
AB - Spatio-temporal forecasting is a cornerstone of logistics, urban planning, and intelligent transportation systems. However, constrained by deployment costs and maintenance resources, sensor networks often lack comprehensive spatial coverage, rendering Forecast Unobserved Node States (FUNS) a critical yet formidable challenge. Conventional models rely on historical observations and typically falter when encountering nodes without prior records. To address this, we redefine the problem as a conditional generation task on spatio-temporal graphs and propose GenST, a framework that introduces Large Language Models (LLMs) as a semantic bridge, leveraging a pre-trained LLM fine-tuned to extract rich semantic features from node descriptions, such as functional zones and road network structures, to compensate for missing spatio-temporal signals. Specifically, we design a two-stage generative architecture: a Spatio-Temporal VAE first compresses spatio-temporal dynamics into a latent space, followed by a Generative Transformer (GenT) that reconstructs the future states of unobserved nodes from noise, guided by multi-modal conditions including semantics, geographic coordinates, and neighborhood contexts. Experiments on six traffic and two non-traffic datasets show GenST significantly outperforms existing baselines in zero-shot prediction tasks, demonstrating the practical potential of semantic-guided generation for mitigating spatio-temporal data sparsity.
ER -